JOURNAL OF THE EXPERIMENTAL ANALYSIS OF BEHAVIOR 1996, 659 185-241 NUMBER I (JANUARY)

ON THE ORIGINS OF NAMING AND OTHER SYMBOLIC BEHAVIOR

PAULINE J. HORNE AND C. FERGUS LOWE
UNIVERSITY OF WALES

We identify naming as the basic unit of verbal behavior, describe the conditions under which it is
learned, and outline its crucial role in the development of stimulus classes and, hence, of symbolic
behavior. Drawing upon B. F. Skinner's functional analysis and the theoretical work of G. H. Mead
and L. S. Vygotsky, we chart how a child, through learning listener behavior and then echoic re-
sponding, learns bidirectional relations between classes of objects or events and his or her own
speaker-listener behavior, thus acquiring naming-a higher order behavioral relation. Once estab-
lished, the bidirectionality incorporated in naming extends across behavior classes such as those
identified by Skinner as the mand, tact, and intraverbal so that each becomes a variant of the name
relation. We indicate how our account informs the specification of rule-governed behavior and pro-
vides the basis for an experimental analysis of symbolic behavior. Furthermore, because naming is
both evoked by, and itself evokes, classes of events it brings about new or emergent behavior such as
that reported in studies of stimulus equivalence. This account is supported by data from a wide range
of match-to-sample studies that also provide evidence that stimulus equivalence in humans is not a
unitary phenomenon but the outcome of a number of different types of naming behavior.
Key words: naming, verbal behavior, language, symbolic behavior, stimulus equivalence, listener
behavior, rule governance, speech for self, consciousness, match to sample, children

Within behavior analysis in recent years observed, it has proved to be enormously dif-
there has been an upsurge of interest in the ficult for the various approaches within psy-
study of human behavior in general and, chology and other disciplines to define what
more particularly, in those complex behavior- a word or a name is and how naming differs
al phenomena that many previously consid- from other forms of behavior (see Quine,
ered to be the exclusive concern of cognitive 1960; Terrace, 1985). Within cognitive psy-
psychology. Language, or verbal behavior, chology, for example, Harnad (1990) has not-
and its interactions with other behavior are ed that a flaw in the dominant paradigm, in
now at the center of a great deal of research, which "the mind is a symbolic system and
and it is increasingly evident that behavior an- cognition is symbol manipulation" (p. 336;
alysts wish to reclaim the high ground of be- see also Fodor, 1975), is that it has a "symbol
havioral complexity and deal with issues such grounding problem" (p. 335; see also Searle,
as word meaning, semantic relations, and 1980): That is, there is no way of relating the
symbolic behavior (e.g., Catania, 1992, pp. symbols in the system to the real world. In
152-156; Dugdale & Lowe, 1990; Hayes & the absence of such a relation, the "symbols"
Hayes, 1992; Lowe, 1983; Sidman, 1990, cannot symbolize or mean anything-a seri-
1992). But to specify, let alone account for, ous flaw by any account.
word meaning and symbolic behavior is not The central aims of the present paper are
at all straightforward. As Premack (1990) has thus ambitious. They are (a) to specify the
basic unit of verbal behavior, or language,
We are grateful to A. Charles Catania for his excellent which we identify as the name relation, and
advice and encouragement and to T. R. Miles for his
helpful comments on an earlier version of the paper. We (b) to show how this behavioral unit is
are indebted to Neil Browning for his work in the pro- learned and comes to symbolize objects and
duction of the figures and to Sue Peet andJudith Brooke events in the real world (cf. Harnad, 1990).
for their efforts in the preparation of the manuscript. We Our account builds upon that of Skinner's
are particularly indebted to Pat Lowe for her inspiration Verbal Behavior (1957), although Skinner,
and editorial skill in making our verbal productions
meaningful. Many of the arguments in this paper were while acknowledging that names differ from
first presented at the May 1994 meeting of the Associa- tacts, did not himself use naming as a techni-
tion for Behavior Analysis in Atlanta. cal term. The term now features prominently
Correspondence and requests for reprints should be in the behavior-analytic literature, but sel-
addressed to either Pauline Horne or Fergus Lowe,
School of Psychology, University of Wales, Bangor, Gwy- dom, if ever, in the context of any behavioral
nedd, LL57 2DG, United Kingdom. specification of what it is. We aim not only to

185
 19383711, 1996, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1901/jeab.1996.65-185 by Capes, Wiley Online Library on [26/05/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
186 PAULINEJ. HORNE and C. FERGUS LOWE
provide a behavioral specification that will course new and had previously been dem-
help to standardize use of the term in the onstrated with many animal species, but Sid-
literature, but also to present a functional man and colleagues found when further test-
analysis of how naming comes about from ing their subjects that entirely new untrained
early infancy onwards, and then, once it is behavioral relations emerged; these relations
established, how it affects, and is in turn af- were not predicted by known laws of learn-
fected by, other behavior. ing. Thus, when presented with the printed
There are problems inherent in using the words (e.g., CAR; lower left, Figure 1), sub-
term naming, laden as it is with connotations jects selected the corresponding picture (e.g.,
from existing usage in behavior analysis, oth- of a car) and when presented with the picture
er science areas (e.g., psycholinguistics, de- they selected the corresponding printed word
velopmental psychology, cognitive psycholo- (lower right, Figure 1). These emergent re-
gy) and, of course, everyday life itself. But lations are shown in Figure 2 as CB and BC,
there are also considerable advantages. An ef- respectively. Following the establishment of
fective behavioral specification of naming 40 trained relations (20 AB and 20 AC), 40
and how it comes about would, if generally new relations (CB and BC) were present
accepted, not only advance behavior analysis when tested. Sidman has described the emer-
but might also foster productive interaction gent performances (CB and BC) as equiva-
with scientists from other traditions who, al- lence relations and has argued that these results
though they may frequently refer to naming, show that stimulus equivalence had been estab-
nevertheless are constrained by the lack of a lished between the three stimulus members
fundamental and specific behavioral under- of each class (i.e., the auditory name stimu-
pinning to its usage. Thus, although we are lus, the corresponding printed word, and the
mindful that our readers will need to hear corresponding picture). Over the two de-
naming as we specify it and will have to ignore cades that have passed since these studies,
misleading connotations from other contexts, their findings have been replicated and ex-
we have nevertheless chosen the term name tended with human subjects of varying ages
to indicate the basic verbal unit. and with differing stimulus modalities and
In recent times many of the innovative types of match-to-sample procedures (see
ideas within behavior analysis about how ver- Saunders & Green, 1992; Sidman, 1994).
bal behavior should be characterized have Describing these behavioral outcomes of
come from researchers working in the area match-to-sample procedures as stimulus
of what has been termed stimulus equivalence. equivalence had very important implications
To understand how this research relates to for subsequent research in this area. In two
linguistic issues it may be helpful to consider key papers, Sidman and colleagues (Sidman
some pioneering work by Murray Sidman, et al., 1982; Sidman & Tailby, 1982) argued
who used "symbolic" match-to-sample pro- that the concept of equivalence as used in
cedures to teach conditional discriminations mathematics could be applied to perfor-
to youths with mental retardation (Sidman, mance on conditional discrimination tasks in
1971; Sidman & Cresson, 1973). The general ways that would bring methodological rigor
procedure is illustrated in Figure 1 which to the definition and identification of "be-
shows that subjects first learned to select (or havioral" equivalence. The three defining
demonstrated that they were already able to characteristics of mathematical equivalence
select) a particular comparison stimulus (e.g., (i.e., reflexivity, symmetry, and transitivity)
picture of a car) upon hearing a dictated were, accordingly, borrowed and applied to
word (e.g., "car"). This performance, which behavioral relations. Thus, according to the
was repeated for many spoken word-picture theorists who have been most influential in
relations, is shown as an AB relation in Figure this area, the property of reflexivity (e.g., A =
2. Subjects were next taught (upper right of A) is inferred from match-to-sample perfor-
Figure 1) to select printed words (e.g., CAR) mance when subjects show generalized iden-
upon hearing the corresponding spoken tity matching; symmetry is inferred when, hav-
words (e.g., "car"), shown as AC in Figure 2. ing trained subjects to select Stimulus B upon
Establishing such conditional relations with presentation of Stimulus A (i.e., an A-B rela-
match-to-sample procedures was not of tion), the subjects proceed without further
 19383711, 1996, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1901/jeab.1996.65-185 by Capes, Wiley Online Library on [26/05/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
NAMING AMD OTHER SYMBOLIC BEHAVIOR 187

TRAINING

I
~ ~ I - IU~

TEST

Fig. 1. An example of a match-to-sample procedure (see Sidman, 1971). This shows a five-key response panel;
during training (top) the sample stimuli are dictated words and the comparison stimuli are visual stimuli (pictures
or words). At the start of each trial, the sample (e.g., the dictated word "car") is presented via tape recorder; touching
the center key then brings on the comparison stimuli on the outer keys. Reinforcers are delivered for selecting the
stimulus that corresponds to the sample (e.g., the picture of a car or the printed word CAR). In test trials (bottom)
the printed words (e.g., CAR) or the picture (e.g., of a car) are presented as samples. Stimulus equivalence is
demonstrated when, in the absence of reinforcement, the corresponding comparison picture is selected when the
printed word is the sample, and vice versa.

training to select A when presented with B er these different behavioral relations under
(i.e., B-A); transitivity is inferred when, having a common name and gave rise to the notion
established A-B and a second relation B-C, that there was a unitary psychological phe-
subjects proceed without further training to nomenon called stimulus equivalence that
select C when presented with A. When sub- could be assayed using match-to-sample pro-
jects show evidence of meeting all three cri- cedures. In addition, what had been a num-
teria with a given set of stimuli (e.g., A, B, ber of different dependent variable measures
and C in the above example), then it is con- on match-to-sample tasks were now viewed as
cluded that all of the behavioral relations measuring a single determining variable-
among the stimuli, even those that have been namely equivalence-that purportedly gives
trained, are in fact equivalence relations and rise to a range of linguistic phenomena that
that the stimuli concerned form an equiva- have puzzled psychologists, psycholinguists,
lence class. and philosophers for many years. Thus, Sid-
The introduction of the mathematically man, for example, has argued that stimulus
based concept of equivalence brought togeth- equivalence is a linguistic prerequisite (Sid-
 19383711, 1996, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1901/jeab.1996.65-185 by Capes, Wiley Online Library on [26/05/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
188 PAULINEJ. HORNE and C. FERGUS LOWE
B lus equivalence is a key variable that trans-
forms nonlinguistic into linguistic behavior,
picures that it is the test of whether relations are se-
mantic, and that it is, indeed, the defining
property of symbolic behavior itself-were to

A
FZ
dictatedIl
/14TI I
AB

{
1

BCOIDCB
be accepted, then this would have major re-
percussions for attempts to explain language.
This, of course, would include Skinner's ac-
count of verbal and rule-governed behavior,
name I which does not incorporate the construct of
4*11 stimulus equivalence. For those wishing to
study linguistic or verbal behavior, what is at
issue here is a specification of the subject
matter itself. When can we say of an utter-
ance, for example, that it is not simply an in-
words stance of operant behavior but that it also has
c properties that enable us to say that it is a
"word" or "name" and that it has "mean-
ing"? For some (e.g., Hayes & Hayes, 1989,
Key: . Trained p. 182), Skinner's definition of verbal behav-
I' 4''' "Emergent" ior as behavior reinforced through the me-
diation of other persons in accordance with
Fig. 2. A schematic representation of Sidman's the practices of the verbal community is too
(1971) equivalence paradigm. The arrows point from broad and fails to distinguish it from any oth-
sample to comparison stimuli. The solid arrows represent er learned social behavior; for these critics it
conditional relations (AB and AC) that were explicitly is equivalence that provides the true touch-
taught. Broken arrows represent conditional relations stone enabling us to specify when behavior is
(BC and CB) that were tested for after the others had
been explicitly taught. verbal. Because it lacked an appreciation of
stimulus equivalence, Skinner's theory of ver-
bal behavior, they argue, ignored such issues
man, 1986, p. 226), and is a determining vari- as meaning, understanding, and reference
able that accounts "both for what people say and, ultimately, failed to grasp the essential
and for their reactions to what other people nature of verbal behavior (Hayes & Hayes,
say. In particular the existence of equivalence 1989, p. 154, 1992, p. 1392). Furthermore,
relations can account for such utterances as both Sidman and Hayes maintain that the ab-
'meaning,' 'symbol,' 'referent,' and 'rule-gov- sence of the concept of equivalence from
erned' " (Sidman, 1992, p. 20). Skinner's account undermines his attempt to
Hayes and colleagues, although adopting a distinguish between contingency-shaped and
different theoretical stance from that of Sid- rule-governed behavior, a distinction many
man, have similarly acknowledged the impor- would see as one of the most important ad-
tance of the phenomenon and have main- vances in the analysis of human behavior (De-
tained that stimulus equivalence "transforms vany, Hayes, & Nelson, 1986, p. 255; Hayes &
nonlinguistic conditional discriminations into Hayes, 1992, p. 1392; Sidman, 1990, pp. 106-
semantic process" (Wulfert & Hayes, 1988, p. 107, 1992, pp. 21-22; Zettde & Hayes, 1982).
126) and is "a kind of working empirical Can it be then, as the "equivalence cri-
model of semantic relations" (Hayes & tique" might suggest, that Skinner's theory of
Hayes, 1992, p. 1387). Still others have pro- verbal and rule-governed behavior is a case
posed that equivalence classes define symbol- of "Hamlet without the Prince"? Before the
ic behavior and that the stimulus equivalence critique and the proposed solutions can be
paradigm provides the basis for an experi- adequately assessed, however, it is necessary
mental analysis of symbolic behavior (e.g., Ca- to examine the notion of stimulus equiva-
tania, 1992, p. 156; Dugdale & Lowe, 1990, p. lence itself and emerging problems in the lit-
115). erature concerning its definition (Hayes,
If these propositions-namely, that stimu- 1989; Pilgrim & Galizio, 1995; Saunders &
 19383711, 1996, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1901/jeab.1996.65-185 by Capes, Wiley Online Library on [26/05/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
NAMIVNG AAM OTHER SYMBOLIC BEHAVIOR 189
Green, 1992) and its origins (Catania, Horne, to not one but a variety of psychological
& Lowe, 1989; Dugdale & Lowe, 1990; Hayes "equivalences."
& Hayes, 1992; Sidman, 1994). There are also Our general approach owes much to Skin-
a number of related issues including the fact ner's Verbal Behavior (1957) but attempts to
that humans, at least those with some verbal extend that account in a number of ways.
skills, readily pass tests of stimulus equiva- Central to our analysis is an emphasis on lis-
lence whereas there are few, if any, clear dem- tener behavior and a conceptualization of the
onstrations of stimulus equivalence in other individual as a speaker-listener within the same
animal species (Dugdale & Lowe, 1990; Hayes skin. This early involvement of the young
& Hayes, 1992; Sidman et al., 1982). child as a listener, first to others' utterances
Attempts to deal with these issues have and then, even more important, to his or her
come from three main theoretical perspec- own, is an aspect of verbal behavior that is to
tives. According to Sidman and colleagues, a large extent not addressed in Skinner's
stimulus equivalence is an unanalyzable prim- book, at least in his account of the basic ver-
itive function that, Sidman suggests, we may bal classes (see Hayes & Hayes, 1989). As he
simply have to accept as a biological "given" himself observed, "Most of my book ... was
(Sidman, 1990). In order to overcome what about the speaker. It contained ... little di-
they see as the shortcomings of Sidman's ac- rect discussion of listening" (Skinner, 1989,
count, Hayes and colleagues (Hayes, 1986; p. 36). Indeed, insofar as his definition of ver-
Hayes & Hayes, 1989) have introduced rela- bal behavior as "being effective only through
tionalframe theory, which views equivalence as the mediation of other persons" is taken lit-
one of several relations arising from a history erally, then the effects of what the speaker
of arbitrarily applicable relational respond- says upon his or her own behavior as a listen-
ing. Lowe (1986)' has put forward a third er, in the absence of any other person, could
view according to which success on match-to- be deemed nonverbal behavior! (but see also
sample tests of stimulus equivalence is attrib- Skinner, 1989, pp. 46-47). Certainly, in all of
utable in large part to subjects' naming and his writings on verbal behavior Skinner says
other verbal behavior, thus explaining the comparatively litfie about speaker-listener re-
lack of success of nonverbal organisms on lations and particularly of how they come
these tests (see also Dugdale & Lowe, 1990; about. In what follows we shall attempt to
Lowe, Horne, & Higson, 1987). Critics of this provide such an account and will argue that
latter approach, however, have rightly ob- it is only through an analysis of both speaker
served that if naming is a prerequisite for and listener behavior that we can establish
passing tests of equivalence, one is left with what counts as an instance of a name, how
the problem of accounting for how naming the latter comes to have meaning and, in
itself comes about and how it gives rise to short, what constitutes linguistic or verbal be-
what appear to be emergent or derived stim- havior distinguishable from other forms of
ulus relations (Catania et al., 1989; Hayes, conditioned responding.
1994; Hayes & Hayes, 1992; Sidman, 1990, To illustrate our general approach and to
1992). In what follows we shall attempt to indicate how speaker-listener relations can ac-
show how naming, although differing from count for a range of emergent behavior in-
other forms of operant or classically condi- cluding stimulus equivalence, consider the
tioned responding, arises out of particular re- different behavioral relations embodied in
lations between such behavior and the envi- the concepts of the tact and the name. Skinner
ronment. We shall also address how naming (1957) defined the tact relation as "a verbal
can enable human subjects to succeed on operant in which a response of a given form
match-to-sample tests for stimulus equiva- is evoked (or at least strengthened) by a par-
lence and how these tests may be passed with ticular object or event," this correspondence
several types of naming behavior, giving rise between the object (e.g., a car) and the par-
ticular form of responding (e.g., saying
1 Lowe, C. F. (1986, May). The role of verbal behavior in "car") being established by generalized re-
the emergence of equivalence classes. Paper presented at the inforcement from the verbal community (pp.
annual meeting of the Association for Behavior Analysis, 81-82). Although tacting is sometimes
Milwaukee. viewed, erroneously, as being the same as
 19383711, 1996, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1901/jeab.1996.65-185 by Capes, Wiley Online Library on [26/05/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
PAUL7NEJ. HORNE and C. FERGUS LOWE
picture of a car), saying ("car"), hearing her
own utterance (/car/)3 and seeing or other-
wise attending to the object again. When, like
66 car 99 the subjects in the Sidman studies, individuals
car with naming skills are trained on match-to-
sample procedures to select first a picture
(e.g., of a car) and later the corresponding
CAR printed word (CAR) upon hearing the name
Fig. 3. The tact relation (Skinner, 19)57) occurs when (Icarl) spoken by the experimenter, this pro-
a response of a given form (e.g., saying "car") is evoked vides ideal conditions for them to learn the
in the presence of a particular object ( e.g., a picture of name themselves for both the picture and the
a car). According to Skinner's accountt, if the stimulus printed word. When, as shown in Figure 4
that evokes the response (e.g., "car") is written or prlnt- (bottom), the picture is then presented as a
ed (e.g., the printed word CAR) then,
the verbal relation is a textualrather thar attacty (pathough sample, the subject either overfly or coverdy
this distinction is not easy to sustain be cause, e.g., a pic- says the name, the hearing of which is dis-
ture may also be drawn or printed). WEien, as in the fig- criminative for orienting to and selecting ei-
ure, the picture and printed word bothievoke the same ther the printed word or the picture from
vocal response (i.e., "car"), this is terme d functional equiv-
aklnce. among the comparisons, whichever is pre-
sented, and vice versa when the sample is the
printed word. Thus subjects may succeed in
naming, it does not, as define d, encompass passing tests of stimulus equivalence by using
a common nane for the stimuli in each class.
any form of listener behavior oin the part of
the speaker, and cannot, by itselIf account for We shall show that other naming strategies,
"emergent" behavior. For exariIple, Figure 3 in particular, such as relating the stimuli by
shows what would have occurre'idhad the sub- means of intraverbal naming, may also be suc-
jects in the studies by Sidman (1 I971)
and Sid- cessful in establishing equivalence classes.
man and Cresson (1973) tacted the stimuliiin Our perspective, although grounded in
Skinner's analysis of verbal behavior, is also
each of the experimenter-design ated stimulus
classes. Each of the stimuli (e.g picture of a informed by accounts of naming and lan-
car and the printed word CAR) would be dis- guage development provided by L. S. Vygot-
criminative for a common resF)onse (in this sky and by the social behaviorist G. H. Mead
instance, the utterance "car"), za relation that (see Blackman, 1991). Our exposition of the
is termed functional equivalence ((Goldiamond, genesis of ontogenetically early forms of lin-
1966; Sidman, 1986). But, given the unidirec- guistic behavior, including the behavioral pre-
tional nature of the tact relati on, there are cursors of naming as well as naming itself,
no grounds for assuming that th is should give draws heavily upon, and is underpinned by, a
rise to bidirectional relations and stimulus great deal of empirical research from the re-
equivalence (i.e., that if given tthe picture of cent literature on language development in
a car, the subject would select tihe word CAR young children. In dealing with normal lan-
and vice versa). Neither tacting nor function- guage development over the first 2 years of
al equivalence in and of itself shtould give rise life up to the point when intraverbal strings
to stimulus equivalence. of a few names have been learned, we aim
The name relation, on the other hand, not to provide an exhaustive description but
which we describe in detail in the next sec- to show the critical steps necessary to bring
tion, involves the speaker respo nding as a lise- about the name relation. It is also important
to note that what we are proposing, with re-
tener to his or her own speakin g. Naming, as
it is learned by the young chiid,2 iS thus a spect both to the order in which these forms
circular relation (shown in Figu:re 4, top) that of behavior occur and the conditions that in-
includes, for example, seeing ari object (e.g., fluence them, is an hypothesis that, although
based on a range of existing findings, re-
2 For purposes of illustration here alnd in the rest of
the paper, we shall take as an examplee a young female 3 To distinguish spoken from heard utterances we shall
child without sensory impairment whc) is learning lan- indicate the former as, for example, "car" and the latter
guage. as /car/.
 19383711, 1996, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1901/jeab.1996.65-185 by Capes, Wiley Online Library on [26/05/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
NAMING AAD OTHER SYMBOLIC BEHAVIOR

says / hears says hears

66car99 Icarl 66 car99 lcarl

sees sees

~~~~~9 Jf
CAR

says hears
66 car99 Icar

V ~~sees F

L_ CAR
Fig. 4. Schematic representation of the name relation and the role of common naming in auditory-vsual match
to sample. The name relation involves speakers responding as listeners to their own speaking and is shown here (e.g.,
top left) as a circular relation between seeing an object (picture of a car), saying ("car"), hearing their own utterance
(Icarl), and seeing or otherwise orienting to the object again. During training on match to sample (cf. Sidman,
1971), shown in the upper displays, when the subject hears the auditory sample stimulus (Icarl), she echoes it (saying
"car") in the presence of the corresponding visual comparison (picture of a car or printed word CAR), and a
common name is thereby established. During subsequent visual-visual test trials (bottom), seeing either the picture
of a car or the printed word CAR as the sample evokes the saying of "car" and the hearing of Icarl, which in turn
occasions selecting either the corresponding picture or printed word, whichever is available as comparison.

quires further detailed empirical validation. THE DEVELOPMENT OF

We shall also show how the present analysis VERBAL BEHAVIOR
informs a range of issues concerning rule- In attempting to show how naming devel-
governed and what has been termed symbolic ops from earlier, prelinguistic behavior, we
behavior. A secondary aim of the paper is to shall relate our account to some of the basic
show how naming skills can account for the verbal classes identified by Skinner (1957), in
success of subjects on tests of stimulus equiv- particular the tact, echoic, intraverbal, and
alence and in a manner that may render the- mand. Skinner defined these classes within
ories of equivalence, and, indeed, the con- the framework of the three-term contingency
struct itself, redundant. (i.e., discriminative stimulus, response, rein-
 19383711, 1996, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1901/jeab.1996.65-185 by Capes, Wiley Online Library on [26/05/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
PAULINEJ. HORNE and C. FERGUS LOWE
forcer) but argued that one of the main dis- behavior to be a crucial precursor to the de-
tinguishing features of verbal behavior was velopment of linguistic behavior and will
that, with the exception of the mand, it was therefore describe it in some detail.
free of specific reinforcing consequences. For The conditions under which listener be-
example, a child's saying "cat" in the pres- havior is established in the young child nor-
ence of a cat could be reinforced by any of a mally include the following: (a) The caregiver
variety of reinforcers (e.g., the caregiver say- or others produce a vocal stimulus, usually an
ing "yes it is a cat," "clever girl," etc.). The object name, in the presence of the object
behavioral modalities were also unspecified and the child; (b) concurrendy, using social
in his account; individuals may learn verbal reinforcement, caregivers teach the child how
behavior regardless of whether they are able to perform conventional behavior in relation
to hear, see, or speak. Although for purposes to the object; (c) rather than her speech sim-
of illustration the present paper focuses upon ply accompanying the child's behavior, the
the development of verbal behavior in a child caregiver's vocal stimulus increasingly pre-
who is without sensory handicap, the main cedes and becomes discriminative for the
principles of our account apply also to people child's performance of this object-related
who are deaf, blind, or have other sensory conventional behavior. These conditions are
impairments, because other responding apart met through a series of behavioral interac-
from vocalizing (e.g., manual signing) can tions between caregiver and infant, as illus-
function as speaker behavior (e.g., Hall & Sund- trated below.
berg, 1987). Again for explanatory purposes Discriminating the speech of others. If the
we have focused upon behavioral interactions speech of others is to provide effective con-
between the child and her primary caregiv- trolling stimuli for the behavior of the young
ers, although we recognize that others (e.g., child, it is essential that this speech and its
siblings, friends, relatives) will also form part component parts (i.e., individual words) be
of the child's verbal community. discriminated by the child. This process be-
gins very early in life. Molfese (1977), for ex-
Listener Behavior ample, reported that when speech sounds are
There is an extensive literature that shows presented to the human neonate, neural ac-
how, before they learn to speak, children tivity is greater in the left cerebral hemi-
learn to listen, as illustrated, for example, by sphere than in the right; when, on the other
a caregiver saying to a child, "give me the hand, music is presented, the converse pat-
cup" and the child doing so. Such behavior, tern of activity occurs. A number of studies
which has been termed listener behavior by using operant procedures have also shown
Skinner (1957, p. 357), arises when the verbal that in the first few days the human infant not
community, of whom we will take the caregiv- only discriminates speech sounds from other
er to be representative, establishes a corre- sounds but also learns to discriminate the
spondence between a vocal or other conven- speech sounds of her native language from
tional stimulus produced by a speaker and those of other languages (Mehler et al., 1988;
behavior evoked in the listener. It includes Morse, 1972). Studies have also shown that at
classically conditioned behavior such as "re- this early stage, speech sounds are more re-
sponses of glands and smooth muscles, me- inforcing for the child's behavior than are
diated by the autonomic nervous system, es- nonspeech sounds (Butterfield & Siperstein,
pecially emotional reactions ... [as well as] 1972), and that her mother's speech is more
... complex skeletal behavior with which the reinforcing than that of a stranger (DeCasper
individual operates upon [the] environment" & Fifer, 1980). Although young infants can
(Skinner, 1957, p. 34). However, according to discriminate a range of phonetic contrasts
Skinner's definition of verbal behavior, listen- made in natural languages, by around 12
er behavior is not verbal except "when the months old their discriminative success be-
listener ... [is] ... to some extent speaking" comes more confined to the subset of speech
(1989, p. 36). Consequendly, this phase of the sounds within their native language (Aslin, Pi-
child's development is given littde considera- soni, & Jusczyk, 1983; Eilers, Wilson, &
tion in his account. In the present analysis, Moore, 1977; Eimas, 1974a, 1974b, 1975; see
however, we consider the learning of listener Gerken, 1994, for recent review).
 19383711, 1996, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1901/jeab.1996.65-185 by Capes, Wiley Online Library on [26/05/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
NAMING AND OTHER SYMBOLIC BEHAVIOR 193

There is a great deal of evidence that shows ing, it enhances the child's subsequent learn-
that children's learning to discriminate oth- ing of listener behavior.
ers' speech is aided by the particular ways in Caregivers indicate the object or event that they
which caregivers speak to them (see Richelle, name. Caregivers also often look at, point to,
1993). In general, caregivers (a) speak slowly and otherwise indicate (e.g., by shaking a
(Snow & Ferguson, 1977); (b) use single rattle) the object or event that they are nam-
names for objects and actions, which are em- ing. By the time they are about 9 months old,
bedded, if at all, in a simplified syntax (e.g., human infants learn to discriminate the be-
"give mummy car"; see Cross, 1977; Newport, havior of the caregiver and to orient toward
Gleitman, & Gleitman, 1977); (c) repeat the object or event at which the caregiver is
names and simple instructions many times looking or pointing; this is learned first for
(Snow, 1977); and (d) often accentuate single objects that are close at hand and later for
names and grammatical boundaries by speak- objects that are at some distance (Butter-
ing with a higher pitch and more exaggerated worth & Cochran, 1980; Butterworth & Grov-
intonational patterns than when addressing er, 1988; Lempers, 1976; Messer, 1978). Stud-
older children and adults (Cooper & Paccia- ies by Baldwin (1991) have shown that the
Cooper, 1980; Mehler, Bertoncini, Barriere, learning of this skill of "joint regard" increas-
&Jassik-Gerschenfeld, 1978; Stern, Spieker, & es the likelihood that the name produced by
MacKain, 1982). These adaptations of adult the caregiver will in turn become discrimi-
verbal behavior, which simplify speech stimuli native for the child's looking only at the ob-
and make them more salient to children ject at which the caregiver is looking or point-
(similar in some respects to stimulus fading ing rather than at other stimuli.
procedures used in operant discrimination The child points at objects and events. After she
studies), are widespread crosscultural fea- has learned to follow the point gestures of
tures of caregivers' speech (Fernald, 1992). others, the child herself begins to point to
In addition, operant techniques have shown objects and is socially reinforced for doing so
that in the case of 4-month-old infants, (Foster, 1979; Lempers, Flavell, & Flavell,
"motherese" is a more effective reinforcer 1977; Leung & Rheingold, 1981; Masur, 1982;
than other forms of adult speech (Fernald,
Murphy, 1978; Murphy & Messer, 1977; Ninio
& Bruner, 1978). Initially, she learns to point
1985). to a particular object, then to the caregiver
If listener behavior to the speech of others and to the object again; later she learns to
is to become fully established, however, it is point to an object while at the same time
necessary notjust for speech stimuli to be dis- looking at the caregiver (Bates, Camaioni, &
criminated (i.e., controlling differential re- Volterra, 1975; Masur, 1983). Her pointing at
sponding of some sort) but when a particular objects is, in turn, discriminative for caregiv-
verbal stimulus is heard (e.g., /where's mama?/) ers' naming of them, which also reinforces
it should be discriminative for conventional the child's behavior. Further naming oppor-
object- or event-related behavior (e.g., look- tunities are provided when the young child
ing at mama). There are a number of ways in picks up or gives objects to caregivers, a com-
which such behavior is brought about. mon behavior in infancy (Bates, Benigni,
Caregivers observe what the child is looking at Bretherton, Camaioni, & Volterra, 1979; Bru-
before they name it. Several studies have shown ner, 1977; Foster, 1979; Gray, 1978).
that caregivers often first observe which ob- Caregivers model conventional behavior. In ad-
ject or event an infant is already looking at dition to pointing at objects when they name
or otherwise engaged with before they speak them, caregivers also pick them up and mod-
about it and reinforce the infant's attention el conventional ways of behaving toward
to it (Collis, 1977; Collis & Schaffer, 1975; them and reinforce the infant's imitative be-
Cross, 1977; Harris, Jones, Brookes, & Grant, havior (Kaye, 1982, pp. 70-83). Infants di-
1986; Harris, Jones, & Grant, 1983; Leung & rectly imitate many of these types of modeled
Rheingold, 1981; Masur, 1982; Murphy, behavior even when the opportunity to imi-
1978). Tomasello and Farrar (1986) have tate is deferred (Meltzoff, 1988; Poulson &
found that when caregivers make such obser- Kymissis, 1988); thus, generalized imitation
vations of the child's behavior prior to speak- enables learning of conventional behavior to
 19383711, 1996, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1901/jeab.1996.65-185 by Capes, Wiley Online Library on [26/05/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
194 PAULINEJ. HORNE and C. FERGUS LOWE
proceed rapidly. If there are kinds of behav- Figure 5 shows what happens when a caregiver
ior that the child is initially unable to imitate, sees a shoe (START 1), points to it, and says
these may be manually prompted and shaped to the child, "where's shoe?" (with an empha-
by caregivers (Kaye, 1982, pp. 80-82; see sis on "shoe"); this request generates for the
Skinner, 1953, pp. 119-122). infant the auditory stimulus /where's shoe?/,
From this point onwards, the child's reac- which is a discriminative stimulus (SD) for
tions to objects and events become increas- looking at or otherwise orienting (R0) to the
ingly conditioned by her verbal community. shoe; this behavior is, in turn, discriminative
Whereas previously she may have investigated for pointing to the shoe (Rx). The child may
quite disparate objects (e.g., a toy truck or a show additional forms of conventional listener
shoe) in very similar ways (e.g., grasping, behavior (e.g., putting the shoe on her foot,
transferring them from hand to hand, raising handing it to the caregiver, etc.) that also pro-
them to her mouth), between 12 and 18 vide reinforcing consequences for her orient-
months of age her actions become differen- ing to the shoe. The caregiver responds by say-
tiated and specific to particular objects (Fen- ing "good girl," which serves to selectively
son, Kagan, Kearsley, & Zelazo, 1976; Hutt, reinforce (SR) the child's listener behavior
1967). For example, whenever she now sees that corresponds with the caregiver's request.
a toy truck she may load it with bricks and It is also possible that the caregiver's request
push it along, and when she sees shoes, she may come to serve as a conditioned stimulus
may put them on and march up and down in (CS) that elicits a range of classically condi-
them, and so on. Increasingly, therefore, ob- tioned responses (CRs). As Skinner (1953) ob-
jects come to occasion particular sequences served, a child may "see X not only when X
of socially conditioned behavior, many of is present, but when any stimulus which has
which are destined to come under the con- frequently accompanied X is present. The din-
trol of the caregiver's utterances during the ner bell not only makes our mouth water, it
establishment of listener relations (Gelman & makes us see food. In the Pavlovian formula
Baillargeon, 1983; Nelson, 1973; Ricciuti, we simply substitute 'seeing food' for 'salivat-
1965; Starkey, 1981; Sugarman, 1982). ing' " (p. 266). In addition, it should be noted
Caregivers reinforce conventional listener behav- that Skinner (1953, pp. 270-276, 1957, p. 158)
ior to vocal stimuli. When the child has learned speculated that there may also be operant vi-
to respond to objects in conventional ways, sualizing or "seeing" of objects when they are
such as pointing at them, picking them up, not present. An experimental demonstration
and giving them to others, the caregiver may, of conditioned "hearing" of stimuli when they
for example, ask her to "bring shoe" and are not present is provided in a study by Hef-
may model appropriate behavior by pointing ferline and Perera (1963).
to, reaching for, and picking up the shoe. This listener behavior sequence, beginning
When the child, imitating the caregiver's ac- at START 1, may be repeated many times un-
tions, reliably brings the shoe in response to til "where's shoe?" alone, without the care-
"bring shoe," the caregiver begins to fade his giver pointing, occasions the child's orienting
or her own point-and-reach gestures while re- to the shoe. Eventually, the sequence may
inforcing the child's correct responses by say- also begin with the child's behavior rather
ing "clever girl," putting the shoe on the than the caregiver's. For instance, when the
child's foot, and so on. Eventually, when the child next sees a shoe (START 2), she may
caregiver simply says, "bring shoe," the child immediately point to it, and when the care-
reliably fetches it and this behavior is rein- giver sees the child's point gesture, he or she
forced. At this stage the caregiver's vocal stim- may initiate the caregiver-speaker/child-listen-
ulus has gained discriminative control of the er behavior sequence (from START 1). Ini-
child's shoe fetching: The child has acquired tially, the listener behavior occasioned by
listener behavior. caregivers' speech may also be partly con-
An illustration of listener behavior: Figure 5 pre- trolled by other contextual stimuli. So, when
sents an example of listener behavior in the asked "where's your shoe?" the infant may
young child; this is the first one of the three point at her shoe when it is in her mother's
main behavioral relations that together make hand but not when it is on the floor. How-
up the name relation depicted in Figure 4. ever, with repetition in different contexts,
 19383711, 1996, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1901/jeab.1996.65-185 by Capes, Wiley Online Library on [26/05/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
NAMING AMD OTHER SYMBOLIC BEHAVIOR 195

Fig. 5. A schematic account of how listener behavior (solid arrows) is learned by the child during interactions
with her caregiver (broken arrows). The sequence begins (START 1) when the caregiver points to and names a shoe,
saying "where's shoe?" (with an emphasis on "shoe"). When the child hears Ishoel, this occasions (SD) listener
behavior, including orienting (R0) and other conventional responding to the shoe such as pointing to it, putting it
on, and so forth (Rx). Hearing /shoe/ may also serve as a conditioned stimulus (CS) evoking conditioned responding
(CRs) such as visualizing a shoe. The caregiver reinforces (SR) the child's conventional listener behavior to "where's
shoe?" by saying "good girl." A similar sequence may also come to be initiated by the child first seeing the shoe
(START 2) and pointing to it (see text). The black arrow constitutes the listener element of the name relation
depicted in Figure 4.

"where's your shoe?" acquires generalized irrespective of its color or size, or whether it
control over the infant's pointing at the shoe be her own shoe, a sports shoe, a sandal, a
regardless of the shoe's location or the par- picture of a shoe in a magazine, or perhaps
ticular speaker (Bates et al., 1975; Luria, even a horseshoe. If she overextends the stim-
1982, pp. 46-47; Reich, 1976; Snyder, Bates, ulus class by pointing to a sock she may be
& Bretherton, 1981). Similarly she learns to told, "no, not shoe, sock," the usual social
respond to the utterance "where's shoe?" (or praise being withheld; conversely, if the care-
just "shoe" alone) spoken by any of a variety giver notices that she underextends the class
of speakers, each of whom produces sounds by not pointing to an item the culture would
of varying pitch and of varying phonetic char- call a "shoe," that item may be indicated to
acteristics (see Luria, 1982). her and named as such (but see Huttenlocher
Listener behavior and stimulus classes. Al- & Smiley, 1987). Thus, as shown in Figure 6,
though Figure 5 depicts one particular shoe, the infant learns when she hears /shoe/ to ori-
in practice caregivers will use the term shoe for ent not just to a particular shoe (left) but to
a variety of shoes and illustrations of shoes. So a class of objects (right), membership of the
when asked "where's the shoe?" the infant class being established by her caregivers who
may come to point at any shoe in her vicinity name each of the different exemplars "shoe."
 19383711, 1996, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1901/jeab.1996.65-185 by Capes, Wiley Online Library on [26/05/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
PAULINEJ. HORNE and C. FERGUS LOWE
hears has generalized to other objects to which the
Ishoel child has already learned orienting listener
behavior (Huttenlocher, 1974; Skinner, 1953,
pp. 93-95, 1957, p. 189). This ensures rapid
generalization to all objects that feature in
the child's existing listener repertoire and
further develops the child's instruction-fol-
lowing repertoire.
sees sees In the course of these early developments,
the child's listening to vocal stimuli becomes
A increasingly consonant with that of her verbal
community. When others utter a given word
or set of words, she orients to a culturally es-
tablished class of objects or events and emits
a culturally defined set of responses. Al-
though it has been shown that other species
(e.g., dolphins, parrots, chimpanzees, sea li-
ons) can be taught to perform at least some
of these listener skills (Herman, Richards, &
Wolz, 1984; Pepperberg, 1983, 1987; Pre-
mack, 1970; Savage-Rumbaugh, 1986; Schus-
terman & Krieger, 1984), unless humans in-
Fig. 6. In learning listener behavior, the child learns tervene, nonhuman animals do not naturally
when she hears /where's shoe?/ to orient not just to one exist in the environmental or social contexts
particular shoe (left) but to a class of objects which her that might enable them to do so.
verbal community names "shoe" (right).
Although the child's listener achievements
are, we maintain, a vital stepping stone in the
In this manner the child begins to learn a se- acquisition of verbal behavior, what distin-
ries of relations, shared with others in her ver- guishes and delimits this stage of develop-
bal community, between auditory stimuli pro- ment is that it is others' speaking that deter-
duced by others and her own responding to mines when and where her listener behavior
classes of stimuli. It is these particular stimulus occurs. We shall next examine how she pro-
classes that she will later learn to name (Har- gresses from being a listener to the verbal
ris, Yeeles, Chasin, & Oakley, 1995). productions of others to becoming a speaker-
Extension of the listener repertoire. There are listener in her own right.
more rapid ways of extending listener behav-
ior than that shown in Figure 5. Once the Echoic Behavior
child learns to orient to a shoe when she Vocal precursors of echoic behavior Although
hears /where's the shoe?!, to a cup when she human infants show a wide range of vocali-
hears /where's the cup ?/, and so on for a variety zations, including the universal consonant-
of objects, the first part of the caregiver's ver- vowel "babbling" patterns (Dale, 1976), vocal
bal stimulus, /where's the / comes to oc-
... approximations to conventional verbal behav-
casion a particular form of orienting behavior ior are not generally acquired until around
(e.g., pointing), whereas the second part de- the end of the first year (Locke, 1980; Oller,
termines the object to which orienting oc- Wieman, Doyle, & Ross, 1976). Using social
curs. The caregiver may now add other forms reinforcement, several studies have shown op-
of listener behavior to the child's repertoire erant conditioning of young infants' vocali-
by, for example, teaching the child to pick up zations (e.g., Rheingold, Gewirtz, & Ross,
a shoe in response to her instruction "pick 1959; Weisberg, 1963), and Routh (1969) has
up shoe" and to pick up a spoon in response demonstrated selective reinforcement of both
to her instruction "pick up spoon." Then, consonant sounds and vowel sounds in in-
when the caregiver says for the first time fants ranging from 2 to 7 months old. In ob-
"pick up cup," the child may pick up the cup, servational studies it has been found that in-
because the listener behavior of picking up fants' babbling matches the tonal qualities
 19383711, 1996, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1901/jeab.1996.65-185 by Capes, Wiley Online Library on [26/05/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
NAMING AND OTHER SYMBOLIC BEHAVIOR 197
produced by their own verbal community Because, unlike Skinner's account, listener
(Weir, 1966), occurs at a higher pitch to behavior is central to our analysis of echoic
mothers as opposed to fathers (Lieberman, behavior, we have in this and subsequent il-
1967), and converges upon a subset of the lustrations attempted to show how the child's
sounds that in combination form words used own speaker and listener behavior may inter-
by their particular linguistic community. 011-
act. This enables us to show how the second
er and Eilers (1988) have shown that infants link of the name relation (see Figure 4) (i.e.,
the relation between the child making an ut-
with severe hearing difficulties, on the other
hand, show very delayed onset of this conso- terance and her hearing of it) is established.
nant-vowel babbling; they also reported that In Figure 7, we have also indicated contextual
the one entirely deaf child in their study nev-
stimuli (e.g., a shoe and the child's pointing
er produced such utterances. Taken together, to the shoe) that occasion the caregiver's ini-
tial verbal response because such contextual
these results indicate that both the onset and
form of infants' babbling are dependent stimuli, although not invariably present in all
upon exposure to the speech sounds of their instances, are a common feature of echoic in-
verbal community. Raising the operant level teractions between the caregiver and the
young child (Harris, Jones, & Grant, 1984).
of such vocalizations is an important step to-
wards the acquisition of speech, given that In Figure 7, the caregiver, seeing the shoe,
the particular syllables produced by infants'says to the child while pointing to the shoe,
babbling tend to be the same ones used in "say shoe, shoe," with vocal emphasis upon
early naming (Vihman, 1986, p. 448). the word "shoe." This generates an auditory
stimulus to which the child responds by ech-
Learning to echo. According to Skinner, the
oing "shoe" (Rv), or some approximation to
early productive repertoire of verbal behavior
consists of separate classes of functional re-
it (e.g., "oo"), a correspondence that is re-
lations. The echoic (Skinner, 1957, p. 55; see
inforced by the caregiver saying "clever girl,"
Catania, in press) is one of the earliest andsmiling, and so on. This relation constitutes
most basic of these, and involves the repro- the echoic as described by Skinner. But, given
duction of the verbal productions of others. that the child has already acquired listener
behavior of the kind described in Figure 5,
That is, infants gradually learn to move their
vocal musculature in such a way as to partially
when the caregiver says "shoe" the resulting
or completely reproduce or "echo" a heard stimulus /shoe/ will also occasion the child's
verbal stimulus; this occurs, for example, listener behavior with respect to the shoe
when caregivers say a given word and rein- (e.g., looking at or orienting to the shoe). In
addition, the child's echoic utterance of
force vocalizations of the infant that approx-
imate the sound of that word. Poulson, Kym- "shoe" generates an auditory stimulus that is
issis, Reeve, Andreatos, and Reeve (1991) also functionally equivalent to the caregiver's
/shoe/ and so may initiate both the listener
have demonstrated vocal imitation in 9- to 13-
month-old infants who were praised when behavior and further utterances of "shoe"
they produced an utterance that matched (i.e., self-echoic utterance); this may in turn
(i.e., approximated) that of an adult vocal lead to yet further repetitions of the echoic
model. In addition, they found that some sequence. At this point the child becomes a
generalized echoic behavior occurred; after speaker-listener with respect to her own ver-
bal stimulus /shoe/.
learning a set of reinforced echoic relations,
the infants imitated novel vocal stimuli in the Maintenance of echoic behavior In her exten-
absence of response-contingent social rein- sive study of the verbal behavior of infants
forcers. Other studies have indicated that and their caregivers, Nelson (1973, p. 47) has
children imitate caregivers' speech a great observed that a striking characteristic of child
deal, particularly in the early stages of lan-
talk is the high frequency of what we have
guage learning (Clark, 1977; Moerk, 1992; described here as echoic and self-echoic be-
Ryan, 1973; Slobin, 1968), and, in addition, havior; she also notes that the functional con-
sequences of the behavior have been little an-
that caregivers imitate their children's utter-
ances (Kaye, 1982; Moerk, 1983). alyzed. The evidence suggests that caregivers'
The kind of interaction that gives rise to behavior can be an important source of re-
echoic behavior is represented in Figure 7. inforcement that establishes and maintains a
 19383711, 1996, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1901/jeab.1996.65-185 by Capes, Wiley Online Library on [26/05/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
PAULINEJ. HORNE and C. FERGUS LOWE

Fig. 7. A schematic account of how echoic behavior is learned by a child who has already acquired listener
behavior to /shoe/. The caregiver (START 1) points to a shoe and says to the child "say shoe, shoe." Hearing /shoe/,
the child herself says "shoe" or some approximation to it (Rv), which is reinforced (SR) by the caregiver saying
"clever girl" and so on. Thereafter, hearing caregivers say "shoe" becomes discriminative (SD) for her echoing
response "shoe." The echoic relation is thus established. But given that the child has previously learned listener
behavior to /shoe/, she now both echoes and responds as a listener to her own vocal stimulus (solid arrows) as well
as to her caregiver's. The sequence may also be initiated by the child (START 2) first seeing the shoe. The black
arrows constitute two of the three main elements of the name relation depicted in Figure 4.

generalized imitative repertoire, although or toddling after it and stroking it (see Skin-
much more research is required to demon- ner, 1957, pp. 57-58).
strate the magnitude and scope of these re- Covert echoic behavior. For the echoic to be
inforcing effects (Kaye, 1982; Moerk, 1992; established as a functional class, the infant's
Poulson et al., 1991). Skinner (1957, p. 164) echoic responses must initially occur overtly
has suggested that the child may also be "au- if they are to be reinforced by the verbal com-
tomatically" reinforced when she duplicates munity. However, it is possible that what be-
the sounds of others. For example, the gins as an overt echoic may come to be emit-
sounds and words uttered by parents may ted at the covert level and become
function as potent classically conditioned increasingly abbreviated in form; this may
stimuli that have strong emotional effects on arise particularly in the case of repetition or
the child so that when she hears her own rep-
lication of these vocal patterns she generates self-echoing (see Clark, 1977, p. 342; Skinner,
stimuli that have similarly strong reinforcing 1957, pp. 76, 141-146). Such covert echoics
consequences. Echoic behavior can also pro- may occasion listener behavior that, as is the
vide other consequences. Thus, for example, case with overt echoics, will have a range of
the child's repeated echoing of the sound reinforcing consequences. Once the child has
/cat/ may sustain her listener behavior of learned generalized echoic behavior (Poul-
looking for the cat; this may then be followed son et al., 1991) and to emit previously
by the positive consequences of her crawling learned overt echoics covertly, new echoics
 19383711, 1996, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1901/jeab.1996.65-185 by Capes, Wiley Online Library on [26/05/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
NAMING AAD OTHER SYMBOLIC BEHAVIOR 199

relation serves to set the conditions for the

says her object itself to enter into direct control of the
child's verbal behavior, thereby occasioning
"&shoe' Ishoel naming.
Naming
As indicated in Figure 7, echoic interac-
tions often begin when caregivers talk about,
and point to, objects the child is looking at,
sees pointing to, or playing with (Harris, 1987; To-
masello, Mannle, & Kruger, 1986; Tomasello
& Todd, 1983; Whitehurst & Vasta, 1975).
AA
The conditions exist, therefore, for these ob-
A jects to enter into functional control of the
child's verbal behavior, and this establishes
A new kinds of verbal relations. One of these,
described by Skinner (1957, p. 83) as the
most important verbal operant, is the tact, in
which a response of a given form is evoked
by a particular object or event. For example,
if a caregiver presents a child with a ball and
the child says "ball," the caregiver reinforces
this correspondence in a variety of ways such
Fig. 8. When the child learns to echo a word spoken as smiling, praising (e.g., saying "good girl"),
by her caregivers (e.g., "shoe"), her own auditory stim- and repeating the child's utterance (e.g.,
ulus Ishoel comes to occasion her looking not just at one
particular shoe but at any object so named by her care- "yes! it's a ball"; see Moerk, 1977, pp. 224-
givers for which she has already acquired listener behav- 235). We have attempted to illustrate in Fig-
ior. ure 9 how, in the course of repeated echoic
interactions of the sort depicted in Figure 7,
may be acquired solely at the covert level of the tact comes about; importantly for the
responding. present account, this also enables us to show
From listening to others to listening to oneself: how the final link of the name relation, the
Although the child can function as a speaker- "closing of the circle," is established. In the
listener only when her speaking is initiated by presence of the child, the caregiver (START
others' vocalizations, once she echoes and 1) points to the shoe and says "shoe"; the
reechoes those stimuli this may help, even if caregiver's verbal stimulus /shoe/ occasions
only briefly, to sustain her listener behavior. the child's looking at the shoe and repeating
This is perhaps the earliest approximation to to herself "shoe," with the result that the
self-instructional behavior. sight of the shoe becomes a frequent ante-
Figure 8 shows that when the child echoes cedent for the utterance "shoe"; this estab-
"shoe," the resulting stimulation occasions lishes the object (shoe) as a discriminative
her looking not just at a single object but at stimulus for the child's own future utterance
exemplars of the class of stimuli for which she of "shoe." From this point onwards, when
has already acquired listener behavior (see the child sees the shoe (START 2), it alone,
Figure 6). When her caregiver says "shoe," without need for the caregiver's speaking, oc-
the child now also says "shoe" in the pres- casions the verbal response "shoe" and its at-
ence of shoes. Although it might appear that tendant listener behavior. It is at this stage
she is able to "name" the object present dur- that we can say the child has learned to name
ing social interactions of the kind we have de- the shoe. Eventually the shoe may also be
scribed in Figure 6, her echoic response "seen' or visualized when it is not present
("shoe") occurs in response, not to the shoe (i.e., conditioned seeing), and this condi-
itself, but to another person saying "shoe." tioned visual stimulus may also give rise to the
However, as we shall next describe, the echoic utterance "shoe" (cf. Huttenlocher & Smiley,
 19383711, 1996, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1901/jeab.1996.65-185 by Capes, Wiley Online Library on [26/05/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
200 PAULINEJ. HORNE and C. FERGUS LOWE

Fig. 9. A schematic account of how naming is learned by a child who has already learned both to echo and to
listen to the auditory stimulus /shoe/. The sequence begins (START 1) when the caregiver points to a shoe and says
"shoe." The auditory stimulus /shoe/ now occasions the child's looking at the shoe while she echoes and reechoes
"shoe." In this way the sight of the shoe becomes a frequent antecedent and then discriminative stimulus (SD) for
her saying "shoe." Thereafter, when the child (START 2) sees the shoe, it alone occasions her saying "shoe" (Rv).
This provides the final link of the name relation depicted in Figure 4-naming is now established. The shoe may in
addition be visualized (CRs) when it is not present (such conditioned seeing being evoked by a reliably accompanying
object, e.g., a sock); the resultant stimulation (SD) may also occasion (gray dotted line) the utterance "shoe" (see
text) .

1987); this relation is indicated by a gray dot- shoe, she says "shoe," hears herself say /shoe/,
ted line in Figure 9. and in response to that stimulus sees the shoe
One part of the total relation described in and otherwise engages with it again. This se-
Figure 9-namely, the relation between the quence may be repeated many times. As Fig-
shoe as a discriminative stimulus and the ver- ure 10 shows, however, hearing herself say
bal response "shoe" (supported by the gen- /shoe/ may occasion her looking not just at a
eralized reinforcement from caregivers) -is particular shoe but at any of the other shoes
sufficient to meet the criterion for Skinner's or pictures of shoes in that listener behavior
tact relation. Although it may be the case, as class (see also Figures 6 and 8). Thus, naming
with the echoic, that tacting can occur with- involves the establishment of bidirectional or
out conventional listener behavior, we have closed loop relations between a class of ob-
shown in the previous sections that the most jects and events and the speaker-listener be-
likely way for a tact-like relation to come havior they occasion.
about is via previously established echoic and This fusion of conventional speaker and lis-
listener relations. This means that when a tener functions establishes a qualitatively new
child sees an object and makes a tact re- bidirectional relation in the child's behavior-
sponse, a verbal stimulus is thus generated al repertoire. From this point on, she does
that occasions her immediate seeing again or not simply respond to an auditory stimulus
reorienting to that object. As shown in Figure originating from others as in preverbal listen-
9, when the child sees an object such as a er behavior, nor does she simply echo vocal
 19383711, 1996, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1901/jeab.1996.65-185 by Capes, Wiley Online Library on [26/05/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
NAMING AND OTHER SYMBOLIC BEHAVIOR 201
ertoire. For example, studies by Huttenlocher
says hears and Smiley (1987) and Harris et al. (1995)
"6shoe" have shown that once infants have learned to
/s,!hoel make a verbal response of a given form in the
presence of a particular object (i.e., a relation
that formally resembles the tact), they almost
invariably also show appropriate listener be-
havior (e.g., if the child can say "dog" in the
presence of a dog, she will also respond ap-
< ~~sees propriately if asked "where's the dog?"). This
linkage at the single name stage between pro-
duction and comprehension in infant behav-
ior has been widely noted in the literature
(Bowerman, 1980; Gruendel, 1977; Rescorla,
1980). The evidence indicates, therefore, that
8008n because listener behavior is so closely tied to
production, studies that explicitly measure in-
fant production repertoires are also provid-
ing an indirect measure of listener behavior
and thus of naming. This is further supported
by the evidence that, once the child learns to
be a speaker-listener with respect to a partic-
ular object, she is also able to name others in
Fig. 10. When the child now sees a shoe to which she the same stimulus class to which she has pre-
has previously oriented as a listener, this evokes the re- viously responded only as a listener (Harris et
sponse "shoe" and, hearing her own auditory stimulus / al., 1995). This would also be expected from
shoel, she orients to any of the shoes in her listener be- our account. For example, we have shown
havior class that are present. She may once again say
"shoe" and again orient to a shoe exemplar and so on. that before she learns to name shoes the
Naming may thus be evoked initially either by seeing a child acquires listener behavior in response
shoe or by hearing /shoe/, and may be reevoked either to others saying "shoe," which is directed to-
by seeing a shoe again or via the self-echoic relation (gray ward a variety of shoes and illustrations of
arrow). In this manner bidirectional relations are estab- shoes; the shoes themselves also come to
lished between a class of objects and the speaker-listener
behavior they occasion. evoke the same conventional behavior as that
occasioned by the vocal stimulus. Given re-
peated instances of the interactions depicted
stimuli, nor does she simply vocalize when in Figure 9, the conventional listener behav-
she sees an object, as the definition of the tact ior directed toward shoes, because it fre-
relation might suggest. Naming has proper- quently precedes the vocal response, itself be-
ties and effects on the rest of behavior that comes discriminative for occasioning the
go far beyond those of tacting, echoing, or name. Thus, whenever she encounters any
listener behavior on their own. As Skinner one of the objects in the previously estab-
(1957, pp. 82, 88) acknowledges, we cannot lished listener class, it occasions a particular
say of a tact that it, for example, represents or pattern of conventional behavior that is com-
reminds us of the stimulus; refers to it; stands mon to those objects, which in turn evokes
for or substitutes for it; specifies or means it. In her utterance "shoe."
sum, the tact relation between a stimulus and But this is only one means by which the
a response is unidirectional and nonsymbolic. name relation continues to evolve over the
On the other hand, as we shall attempt to lifetime of the individual. For example, the
show, the name relation as outlined here has child may extend a name (e.g., "shoe") be-
all of these defining characteristics of sym- yond the original stimulus to new stimuli that
bolic behavior (Sinha, 1988). either physically resemble it (e.g., boots) or
Indeed, there is evidence that casts doubt have reliably accompanied it in the past (e.g.,
on whether tacts, as opposed to names, fea- socks). Extension of the name relation (e.g.,
ture very much in the child's early verbal rep- "shoe") may also occur when a novel object
 19383711, 1996, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1901/jeab.1996.65-185 by Capes, Wiley Online Library on [26/05/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
202 PAULINEJ. HORNE and C. FERGUS LOWE
evokes behavior (e.g., putting her foot in a echo will determine what listener behavior
box) that closely resembles listener behavior can become incorporated into name rela-
already embodied in a name relation; the tions, listener skills (or comprehension)
common behavior evokes the name ("shoe") should accordingly outpace naming skills un-
in both instances, and the novel object is now til the echoic repertoire is well advanced.
so named. The name relation may therefore This prediction is supported by many studies
be extended on the basis of the common that have shown that, initially, children's
function as well as common physical features learning of listener behavior to words is much
between a novel object and the original ref- faster than their production of those words
erent (Lewis, 1936, pp. 197-199). (Benedict, 1979), so that on average, by the
Some extensions, as in the examples above, time 60 listener relations have been learned
involve idiosyncratic generalizations that are the child has produced only 10 words (Ben-
at odds with the way the verbal community edict, 1979; Goldin-Meadow, Seligman, &
uses a particular name. Idiosyncratic exten- Gelman, 1976; Greenfield & Smith, 1976, pp.
sions of this kind may be corrected by care- 219-220; Snyder et al., 1981). Later, however,
givers. Thus in the case of overextension as the naming and the echoic repertoires in-
(overgeneralization), the caregiver may, for crease, the discrepancy between listener and
example, say "no, that's not a dog, that's a production skills diminishes (Anisfeld, 1984,
cat." On the other hand, if the child says p. 82).
"cup" only in response to seeing a particular The acquisition of an echoic repertoire
cup, the caregiver will point out other cup may also have an important bearing upon the
exemplars and establish the name to those common phenomenon described in the lit-
also. In this way a balance is struck between erature as the naming explosion. This refers to
the child's tendency to idiosyncratically gen- the very rapid acceleration in name acquisi-
eralize the name on the basis of structural tion that occurs after approximately the first
and functional properties already encom- 10 to 20 names have been learned, normally
passed by it on the one hand, and, on the when children are around 18 months old. Ac-
other, the continuing efforts of the verbal cording to the present account, it may not be
community that keep the name relation with- possible for many new names to be acquired
in the bounds of the socially shared (Vygot- until a critical number of echoic relations,
sky, 1934/1987, pp. 255-285). with differing phonetic characteristics, have
Echoic behavior as a determinant of naming. been learned. As the number of these echoic
According to our account, the echoic reper- relations in the repertoire increases, the com-
toire constitutes a critical link in the devel- binatorial possibilities for producing more
opment of the name relation because it de- name utterances rises exponentially. A fur-
termines whether a listener relation can ther determinant of the name explosion is
become a speaker-listener relation. It should considered in the next section.
also be noted that in early word learning, Naming as a new higher order relation. In the
when the echoic repertoire is very limited, in- interactions between child and caregiver de-
fants can echo only those names (and then scribed so far, we have noted that joint visual
only partially) that have a phonetic element attention to objects is often established by
in common with their existing echoic reper- means of cues provided by the caregiver such
toire (cf. Skinner, 1957, pp. 59-60). What the as pointing or phrases such as "where's the
child can echo will thus determine what she ... or "that is a ....." Although early name
can name. In support of this account, several relations may be learned piecemeal (see Fig-
studies have shown that children's early ver- ures 5, 7, and 9) with listener, echoic, and tact
bal productions that approximate caregivers' components established separately, with rep-
names are heavily dependent upon the small etitions of the interactions shown in Figure 9
set of speech sounds they can already articu- the cues of the caregiver's naming of and
late either singly or in combination (Fergu- pointing at a new object come to be sufficient
son & Farwell, 1975; Ferguson, Peizer, & on their own to evoke the full sequence of
Weeks, 1973; Menn, 1976; Schwartz & Leon- behavior that makes up the name relation.
ard, 1982). Nelson and Bonvillian (1973), for example,
Because the vocal stimuli the child can have reported that children 18 months and
 19383711, 1996, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1901/jeab.1996.65-185 by Capes, Wiley Online Library on [26/05/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
NAMING AND OTHER SYMBOLIC BEHAVIOR 203
older can learn new object names after an drink, a cuddle, or play) that may also serve
adult has named the objects in their presence as a rich source of reinforcement that will
only once or twice. Thus, the name relation maintain the naming behavior that preceded
is established as a higher order behavioral relation it (Huttenlocher & Smiley, 1987). These and
(see also Catania, in press) that short-circuits other effects of naming (see below) serve to
earlier means of constructing it. As with the maintain it at high strength and begin to free
echoic, higher order naming may become in- it to some extent from the direct reinforcing
creasingly covert and abbreviated in form and practices of the verbal community. This does
may indeed be learned at the covert level of not mean, of course, that the verbal com-
responding, this being reinforced and main- munity then ceases to play a decisive role. On
tained by a range of consequences (see Skin- the contrary, caregivers and others continue
ner, 1957, pp. 141-146). to exert a powerful influence on its further
Maintenance of naming. Initially, the name development, maintenance, and generaliza-
relation in young children is supported by tion (Harris, 1992; Moerk, 1983).
caregivers providing a range of generalized The development of naming. From the time
reinforcers, such as smiling, praising, and re- she first names an object or event, the child's
peating what the child has said. Studies have name relations continue to change and de-
shown that thereafter the frequency of exter- velop throughout her life. For any particular
nal reinforcers from caregivers and others name, additional stimulus events come to oc-
simply for correctly naming objects declines casion her verbal utterance, and there are
markedly (Moerk, 1977, p. 265). However, concomitant changes in the listener behavior
when the child names an object in the pres- that the utterance evokes. This is the means
ence of a caregiver, she almost invariably by which socially shared listener behavior as
evokes the latter's attention both to herself well as idiosyncratic listener behavior are es-
and to the object that she has just named and tablished (see Mead, 1934, pp. 135-226; Vy-
with which she is engaged; this attention is a gotsky, 1934/1987, pp. 255-285). We have
rich source of reinforcement for the child's tried to capture this process in Figure 11,
naming, as is the joint activity with the named which shows what happens when a child, who
object that often ensues (Prorok, 1980). has previously learned to say "dog" only in
Important sources of reinforcement, how- the presence of toy dogs and pictures in
ever, are increasingly embodied within the books and to point to them upon hearing
name relation itself. For example, the rein- "where's the dog?" encounters a real dog for
forcement provided by the conditioned stim- the first time. When the child first sees the
uli, operant or Pavlovian, that naming pro- dog (A), the visual resemblance between the
duces and by the listener behavior that previously seen objects and the real dog is
naming occasions (together with its conse- sufficient to evoke, via stimulus generaliza-
quences) may be sufficient to maintain nam- tion, the child's verbal response "dog" and
ing behavior at strength (see Skinner, 1957, her pointing to the dog. She now sees a dog
pp. 86, 163-166, 438-452). Thus a child who that moves, and she hears the sounds of pant-
emits the name "mama" in her mother's ab- ing and whining that are characteristic of real
sence may "see" her, "hear" her, "smell" her dogs. These new properties, as well as the
special scent, and "feel" her comforting previously learned visual features, come to be
touch. Naming that, via conditioning, gener- discriminative for her saying "dog" in future
ates such effects is reinforced because it does encounters with dogs. Her saying "dog" gen-
so. The verbal stimulus may also serve as a erates the auditory stimulus Idogl that in turn
discriminative stimulus for a range of listener evokes conditioned visualization of previously
behavior with a variety of consequences. For encountered pictures and toy dogs but also
example, the child's saying the word "mama" now occasions visual reorientation to the real
may set the occasion for her to visually or oth- dog that is present. As she repeats the name
erwise search the environment until she finds "dog," further pairings of /dog/ with seeing
her mother-a potent reinforcer for that and hearing a real-life dog may result in the
naming behavior. Once "mama" is found verbal stimulus evoking thereafter not just
and held in attention, other behavior can be conditioned seeing but also conditioned
initiated (e.g., her mother may provide a hearing of a dog (shown in B), each of which,
 19383711, 1996, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1901/jeab.1996.65-185 by Capes, Wiley Online Library on [26/05/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
204 PAULINEJ. HORNE and C. FERGUS LOWE

A D
says hears rs ~~ hers

Idogl
I Mim

Fig. 11. Ongoing multimodal development of the name relation: The child in this example has previously learned
to say "dog" when seeing pictures of dogs and toy dogs. She next learns to say "dog" also when she sees a real dog;
sees it move and hears it pant (A), touches its wet nose (B), pats its back, smells and feels its coat (C), and hears it
bark (D). The name relation may now be evoked by any one, or some combination of, these new stimuli. In addition
(as indicated by the inner gray arrows of name relations A, B, C and D), the auditory stimulus /dog/ comes to
occasion conditioned seeing, feeling, smelling, and hearing of dogs, which may, in turn, evoke saying "dog" and so
on.

even in the absence of a real dog, may later ing, accumulated from previous encounters
set the occasion for further emissions of the with dogs, listener behavior that itself is
name. Similarly, when the child touches the changed by each further encounter with dogs
dog's wet nose (B), pats its back, smells and or any of their attributes and the conse-
feels its coat (C), and then hears it bark (D), quences of these interactions. Thus, the stim-
each of these new events join the previously ulus /dog/ comes to evoke a myriad of sights,
encountered stimuli as potential discrimina- sounds, tactile sensations, and smells of dogs
tive stimuli for the emission of the verbal re- and conditioned affective responding.
sponse "dog" in the future. In addition, the Naming and functional stimulus classes. The
auditory stimulus Idog! will henceforth give verbal community instructs the child not only
rise to new listener behavior including, for which objects with similar appearance have
example, touching, patting, and stroking, as the same name but also assigns common class
well as conditioned seeing, feeling, smelling, names to objects that, although they may
and hearing of dogs and properties of dogs. have little in common physically, serve a
Figure 12 shows in detail how the child's shared cultural function, that is, are function-
name relation, "dog," alters in the course of ally equivalent (Mead, 1934, pp. 83-85;
her interactions with the real dog. This name Brown, 1958, p. 85). The name "chair," for
relation will continue to develop in the instance, is used in this culture for a range of
course of further encounters with other objects that are of varied shape, color, and
breeds of dogs, other features of individual size and made of differing materials. Having
dogs, and other illustrations of dogs. And on learned to use "chair" in the single case of
each new occasion the name "dog" is ut- her bouncing chair, the child may not extend
tered, it will carry with it the generalized lis- the name to other chairs until the conven-
tener behavior, including affective respond- tional class is established by caregivers who
 19383711, 1996, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1901/jeab.1996.65-185 by Capes, Wiley Online Library on [26/05/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
NAMING AND OTHER SYMBOLIC BEHAVIOR 205
it is a "chair." It should be noted that this
extension of the name has not been directly
says hears trained or reinforced but occurs "spontane-
ously" via common listener behavior (i.e., sit-
"dog99 Idogl ting) that gives rise to the verbal response
"chair." In this way there is a blending of cul-
tural and personal factors to further extend
the stimulus class that bears the common
name.
At a later stage, the child may, for example,
be taught the name "furniture" for her var-
ious items of doll furniture for which she al-
ready has other names (e.g., "chairs," "ta-
D
bles," "beds," etc.). Although at one level of
classification chairs are objects that people or
41U
dolls sit on, at another more general level
chairs are furniture items to be arranged and
sees OOther listener
behavior to dog rearranged with other such items and moved
hears pointsto from room to room or from house to house
feels! touches nose
sesent amt and so on. When the child calls chairs "fur-
smells ben ds:tdback
strokes
sopwan niture," she moves to a more general level of
functional equivalence. In this way, naming,
cuddles
Fig. 12. Schematic representation of changes in the
incorporating as it does bidirectional rela-
name relation across modalities resulting from the inter- tions between a single verbal response and a
actions depicted in Figure 11. When the child now sees, class of objects or events, can establish and
hears, feels, or smells a dog, she may say "dog." When maintain more than one level of functional
in turn she hears her auditory stimulus /dog/ she may equivalence.
orient to a dog, and point to it, touch its nose, pat it, and
so on. In addition, the auditory stimulus /dog/ may evoke
Naming and emergent behaviotr The fact that
(lower gray arrow) conditioned seeing, hearing, feeling, names relate to classes of objects has pro-
and smelling of dogs (CRs). The echoic relation between found implications for the development of
her own auditory stimulus Idogl and her saying "dog" new or untrained behavior. Figure 13 (left)
(upper gray arrow) may reevoke this conditioned re- shows, for example, a child who has learned
sponding and occasion further revolutions of the name
relation as a whole. to use the name "furniture" for a range of
items of doll furniture and who, in addition,
has been shown how to put some of these
point to other "chairs," naming them as items (e.g., chairs and chests of drawers), but
such. The child's acquisition of the name of not others, into a toy van. If now, as shown in
a conventional stimulus class such as "chair" Figure 13 (right), the caregiver instructs her
confers major functional benefits. For exam- to "put the furniture into the van," she may
ple, having been taught to sit on what care- scan the toys repeating to herself "furniture"
givers call a "chair," the child adopts the cul- and extend to all the furniture items the
tural function of chairs in general when she same listener behavior that has been estab-
adopts the name and extends it, together lished previouslyjust for the chairs and chests
with associated listener behavior, to new ex- of drawers; that is, she will put all the furni-
emplars. So when told that a new and unusu- ture and not just the latter items into the van.
al object is a "chair," the child will immedi- There is thus no need for the caregiver to
ately be able to respond (i.e., by sitting on it) reinforce the target behavior with each indi-
in a way that is both conventionally accept- vidual object; it is sufficient to establish it with
able and intrinsically reinforcing. These cul- some of the exemplars and then use the
turally defined functions thus become for her name "furniture" to effect gpneralization to
an integral part of her listener behavior to the remaining items in the furniture class. We
the name "chair." Indeed, when she sees have here, in other words, new responding
someone sitting on an object or finds that she directed at all of these additional items; the
can sit comfortably on it, she may declare that child treats each in a similar way, that is, as if
 19383711, 1996, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1901/jeab.1996.65-185 by Capes, Wiley Online Library on [26/05/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
PAULINEJ. HORNE and C. FERGUS LOWE

says 7eo> hears says hears

ufiurniture" Ifumiturel "furnirtre" Ifurniturel

sees sees

KeO
- LooT at dbet
-uu3'o PhIs In vwa(ttslned)
ftommftw)k*
4,-
Fig. 13. Schematic representation of how naming brings about new behavior. A child is taught to name individual
items of dolls' furniture (e.g., a chair, chest of drawers, lamp, television) as "furniture." Independently of this she is
trained to load some of them (e.g., chair and chest of drawers) into a toy van (left). Subsequently, when instructed
to "put the furniture in the van," she may show the untrained behavior of putting the lamp and television into the
van (right).

each were functionally equivalent. Other con- 14, this comes about because the pair of ob-
ventional behavior pertaining to the class jects occasions her saying "two" which in
"furniture" may be rapidly learned in the turn evokes the previously established listener
same manner. This brings us to the question behavior of selecting the numeral 2. In a sim-
of how naming can give rise to emergent or ilar manner, when later shown the numeral 2
derived relations, such as those observed in on the sample page, she may be able to select
studies of stimulus equivalence. the pair configuration from among groups of
If taught a common name for members of differing numbers of objects on the compar-
a class of physically different stimuli, a child ison page. Later she may learn that the print-
may, when presented with a single exemplar ed word TWO is also named "two" and when
of the class, select other class members with- then presented with TWO may match it var-
out having ever previously been directly iously to the numeral 2 or to two objects, in
trained to do so (see Eikeseth & Smith, appropriate choice contexts.
1992). For example, a child may be taught to Thus, as shown in Figure 14, by teaching
name as "two" pairs of objects (e.g., two dots, the child to use the common name "two" for
two apples, etc.) and, independently, to name each of the three stimuli, up to six new be-
and select as "two" the numeral 2. When, as havioral relations may emerge, apparently
part of a matching game, she is subsequently spontaneously, but in fact through the listen-
shown a pair of objects on one page of a book er behavior engendered by the name. Evi-
and asked to "pick the one that goes with dence that new behavioral relations can be
them," she may be able, without further established in this manner comes from a
training, to select the numeral 2 from an ad- study by Eikeseth and Smith (1992) with au-
jacent page illustrating different numerals. tistic children (see pp. 224-225), although
According to our account, as shown in Figure further research is needed to provide a sys-
 19383711, 1996, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1901/jeab.1996.65-185 by Capes, Wiley Online Library on [26/05/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
NAMING AND OTHER SYMBOLIC BEHAVIOR

caregivers ostensibly teach the child only con-

ventional listener behavior (e.g., orienting to
says hears and picking up a shoe in response to the ut-
"two" Itwol terance "where's shoe?"), she will neverthe-
less also exhibit the corresponding speaker
behavior (saying "shoe" in response to see-
ing the shoe). Likewise, when only speaker
behavior is ostensibly taught (e.g., when the
caregiver points to a dog and says "dog," the
child repeats the utterance and learns to say
"dog" herself when she sees it), the child also
acquires listener behavior (so that when next
asked "where's dog?" she orients and points
sees to it). Second, we have shown that once the
higher order name relation has been learned
by the child, there may be no need for the
verbal community to provide reinforcement
to establish appropriate speaker and listener
behavior; it may be sufficient, for example,
for caregivers merely to point to and utter the
name of a novel object for the full name re-
lation, incorporating both speaker and listen-
er behavior, to be established. A third crucial
Fig. 14. When a child is taught to name as "two" the
numeral 2, two dots, and the printed word TWO (solid feature of a name is that it is used by the
arrows), up to six new behavioral relations (broken ar- verbal community, and hence the child, not
rows) may result. Whichever of these stimuli is presented for a particular stimulus but for conventional
to her she will say "two" and hear Itwol; this will evoke classes of objects and events. A child learns
her orienting towards and selecting from the available to say "shoe" when she sees one of many
comparison stimuli any one or more of those stimuli that
she has previously named "two." kinds of shoe, from any of a multitude of per-
spectives, in any of a multitude of contexts.
As a consequence, new listener behavior
tematic investigation of this kind of emergent trained with just one or two members of the
behavior. There are many other possible ex- named class may come to be occasioned by
amples of such stimulus classes, the child any one or all of the others, and training a
learning in a similar way the interchangeabil- common name for a number of different ob-
ity in some contexts of particular object class- jects results in the child's treating them in-
es (e.g., dog), particular spoken words (e.g., terchangeably in some contexts.
"dog"), and particular arrangements of To summarize, we can now define naming
printed letters (e.g., DOG). Any of these stim- as a higher order bidirectional behavioral re-
uli may occasion the same name so that the lation that (a) combines conventional speak-
same listener behavior (e.g., selecting) may er and listener behavior within the individual,
be directed towards other members of the (b) does not require reinforcement of both
same stimulus class. Such behavior has been speaker and listener behavior for each new
termed stimulus equivalence, and we have name to be established, and (c) relates to
shown how one very basic form of naming- classes of objects and events. This definition
that is, common naming-can bring it about. has clear implications for attempts to deter-
We shall address others below. mine experimentally whether or not subjects
Towards a definition of naming. According to can name. One possible test would involve
our account, naming is a higher order bidi- the training of speaker behavior in relation
rectional behavioral relation that combines to a novel stimulus and then testing for the
conventional speaker and listener functions presence of conventional listener behavior.
so that the presence of either one presup- This could, for example, be achieved as fol-
poses the other. Thus, when higher order lows. First it would be necessary to ensure
naming skills have been established, even if that conventional listener behavior had been
 19383711, 1996, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1901/jeab.1996.65-185 by Capes, Wiley Online Library on [26/05/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
208 PAULIJNEJ. HORNE and C. FERGUS LOWE
established with other stimuli (e.g., looking match to sample, in the naming procedure
or pointing at the correct object when the stimuli from the same class are not presented
caregiver says "where's X?"). The caregiver together but are presented as individual ex-
would then point to a novel object (e.g., a toy emplars named individually by the experi-
giraffe), say "giraffe," and reinforce the sub- menter. In the procedure we have outlined
ject's saying "giraffe" when the latter was pre- here, the subject's speaker behavior would be
sented; other vocal responses could be estab- reinforced during training, but even this
lished with other novel objects in a similar might not be necessary.
fashion. The test for naming would consist of These naming tests, like tests of stimulus
presenting all the objects together and asking equivalence, all involve what might be termed
the subject, "where's the giraffe?" If the sub- emergent or derived relations. That is, we direct
ject had learned the name, as distinct from reinforcement or training at speaker relations
mere speaker behavior or tacting, she should alone but obtain new or additional listener
proceed to show the conventional listener be- relations. In the case of the stimulus class
havior of looking, or pointing, at the toy gi- tests, we confine training to a set of object-
raffe. word relations but get many new kinds of ob-
A more demanding test would be to follow ject-object listener behavior. In both cases, in-
a similar procedure to that outlined above sofar as the new behavior has not been
but, when the caregiver named the objects in explicitly reinforced during the testing pro-
the presence of the subject, provide no rein- cedures, it might be said that it is emergent
forcement. Given that the higher order name or derived. These terms are most often used,
relation had already been learned by the sub- however, to describe behavior for which the
ject with other objects and events, this pro- controlling variables are unknown (see Mead,
cedure might nevertheless be sufficient to es- 1934, p. 198); hence, they frequently occur
tablish both conventional speaker and in the literature on stimulus equivalence. Our
listener behavior for novel objects and events. account of the emergent phenomena in the
A yet more stringent test would be one that above cases, on the other hand, shows that
capitalized upon the fact that naming is con- they all involve learned behavior and rela-
cerned with stimulus classes. Having taught tions that are established through naming. Al-
either conventional speaker or listener be- though naming is responsible for bringing
havior in the presence of a given stimulus into the repertoire a great many new behav-
(e.g., a shoe) in the manner outlined above, ioral relations, it is itself learned behavior,
either with or without reinforcement, it and its operations can be understood
should be possible, if naming has been estab- through behavior analysis. To the extent that
lished, to extend it to other physically differ- we can give an account of naming and its ef-
ent exemplars of shoe by the caregiver simply fects on other behavior, the concepts of emer-
pointing to each of them and saying "shoe," gent and derived may in this context become
with the subject's correct speaker behavior redundant.
being reinforced throughout. The same Naming through other modalities. The present
could be done for other object classes account, although based upon behavioral in-
(clothes, furniture, toys). If the subject has teractions between caregivers and children
learned the name "shoe," then, in the pres- who are not sensory impaired, should also be
ence of exemplars of shoes among other ob- applicable to those who do have sensory im-
jects, when the caregiver presents one of the pairments (e.g., those who are deaf or blind).
shoes saying "see this, where are the others?" In order to learn the name relation, such
the child should be able, via her own overt children require a sensory modality through
or covert utterance of the name, to select oth- which they may learn (a) to discriminate ar-
ers in the class. This final part of the test, bitrary but conventional stimuli provided by
involving as it would the presentation of a their verbal community (which might, in the
sample and the requirement that the subject case of deaf children be manual signing, or
choose the appropriate comparison stimuli, in the case of blind children, speech), (b)
bears a close resemblance to match-to-sample some of the conventional functions of objects
tests for stimulus equivalence. The principal and to perform these forms of behavior in
difference lies in the training given. Unlike response to the corresponding conventional
 19383711, 1996, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1901/jeab.1996.65-185 by Capes, Wiley Online Library on [26/05/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
NAMING AND OTHER SYMBOLIC BEHAVIOR 209

stimuli (i.e., listener behavior), and (c) to re- 244; Skinner, 1957, p. 56; Slobin, 1968; White-
produce (echo) those stimuli themselves hurst & Vasta, 1975). However, in what
(e.g., by signing or speaking) while respond- follows we shall focus upon intraverbal rela-
ing to the corresponding objects. In this way tions between names.
these children can learn to emit arbitrary but Bidirectional relations between names. With
conventional behavior that evokes their own self-echoic repetition, which is a common fea-
conventional listener behavior as well as that ture of children's verbal play (Moerk, 1992),
of others in their verbal community who the relation between the individual name
share the same conventions. Under such con- components in an intraverbal is easily re-
ditions, naming should be effectively estab- versed (Clark, 1977). Thus, the child's repe-
lished. Indeed the evidence shows that young tition of the intraverbal "knife fork" yields
children who are deaf or blind learn to sign "knife fork knife fork knife fork" and, be-
or speak, respectively, and thus to name, at cause "fork" frequently precedes "knife,"
rates similar to those of children without sen- the child may come to say "fork knife." Clark
sory impairment (Folven & Bonvillian, 1991; and van Buren (1973, p. 85), for example,
Mulford, 1988; Pettito, 1992). report that a 2-year-old child learned in this
fashion to make the following utterances:
Intraverbal Naming "towel wet towel," "noisy man noisy," "back
The effectiveness of naming in bringing pin back," and "soap want it soap." The in-
about new or emergent behavior is greadly en- traverbal relation is therefore not a rigid uni-
hanced as soon as the child begins to use directional sequence of behavioral compo-
names not just singly but in combination. nents but may be bidirectional. If additional
Once children at around the age of 18 terms are added to the sequence, then many
months have acquired a repertoire of, on av- new combinations may emerge. For example,
erage, 50 approximations to single conven- if the child comes to self-echo "fork spoon"
tional verbal responses they begin to combine in addition to "fork knife," then when she
these, initially to form response pairs (Anis- next responds to the stimulus /fork/, it may
feld, 1984). For example, the caregiver may be to say either "spoon" or "knife," because
often repeat the words "big teddy" to the both have in the past regularly followed this
child and the child may echo these words; in stimulus (cf. Skinner, 1957, pp. 71-76). With
this way, the verbal stimulus /big/ comes to still further echoic repetitions, the end result
reliably occasion the child saying "teddy." may be sequences such as "fork knife spoon"
When the caregiver says "big ... .?" the child or "fork spoon knife" and so on. Thus many
says "teddy" and is praised by the caregiver possible combinations of the constituent
for doing so (Goodwin, 1985; Greenfield, terms of an intraverbal may be produced by
Reilly, Leaper, & Baker, 1985; Harris, 1987). the child and, depending in part upon their
This form of verbal behavior has been proximity to conventional usage, some of
termed the intraverbal by Skinner (1957, p. these may be reinforced by the verbal com-
71) and it is characterized by a relation in munity.
which, like the echoic, a verbal stimulus is the How intraverbals can reflect conjunctions of ob-
occasion on which a particular verbal re- jects and events. The child learns intraverbal
sponse receives generalized reinforcement. behavior not only through echoing the verbal
However, we propose that, as in the case of behavior of caregivers and others but also
the echoic and the tact, the intraverbal can through naming successively two or more
exist in two forms, one with and one without items, or properties of items, or events that
conventional listener behavior. There is in- tend to coexist in her environment (Skinner,
deed evidence (a) that children do learn 1957, p. 75). A child who has learned to
from their caregivers sequences of vocal ut- name a fork and a spoon individually may
terances or "words" (e.g., simple repetitions also come to say "spoon fork" or "fork
of adult speech such as nursery rhymes) that spoon" when she regularly sees the spoon
are initially devoid of conventional listener and fork together at meal times. With self-
behavior, and (b) that this may be important echoic repetition the two names may become
for the subsequent development of intraver- bidirectionally related so that when either is
bal naming (Clark, 1977; Moerk, 1977, p. emitted it occasions the other. For example,
 19383711, 1996, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1901/jeab.1996.65-185 by Capes, Wiley Online Library on [26/05/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
210 PAULlNEJ HORNE and C. FERGUS LOWE
upon seeing a spoon the child may say tendant listener behavior. Similarly, if the
"spoon fork" and, in the absence of the fork, child already has learned the names "fire,"
search for it (Skinner, 1957, pp. 75-76). This "cooker," and "poker," the caregiver can go
type of intraverbal relation has particular rel- on to teach her "hot fire," "hot cooker," and
evance for studies of stimulus equivalence "hot poker," all of which extend the listener
(see pp. 218-221, Intraverbal Naming and Stim- behavior to the name "hot" and also to the
ulus Equivalence). names "fire," "cooker," and "poker." There
Interchange of listener behavior across intraver- is thus no need for the child to learn from
bal names. Irrespective of how intraverbals are experience how all of these objects can harm
acquired, their emergence heralds important her, because the listener behavior established
new developments in name relations. As we to just one of them occasions avoidance of all
have already shown, a name becomes increas- the rest. This intraverbal means of transfer-
ingly generalized in terms of the stimuli that ring listener behavior and establishing func-
occasion it and the listener behavior it sets in tional equivalence between physically differ-
train, and included among these stimuli that ent stimuli has considerable reinforcing
occasion it may now be the verbal stimulation consequences both for the child herself and
arising from another name with which it has for the caregiver.
been intraverbally related. The complexity of In the above example, the word kettle in its
the relation between words within intraverbal intraverbal context (i.e., "hot kettie") evokes
sequences has been recognized by Skinner as listener behavior (e.g., avoidance and condi-
follows: "Many different [verbal] responses tioned emotional responses) additional to
are brought under the control of a given that occurring when it is used on its own. On
stimulus word, and many different stimulus the other hand, "hot kettie" also eliminates
words are placed in control of a single [ver- some of the listener behavior that "kettie"
bal] response" (1957, p. 74). What is not ac- alone might evoke (e.g., touching and play-
knowledged in this passage, however, is that ing with it). Because individual names be-
when names are reliably linked within an in- come increasingly generalized and extended
traverbal sequence, listener behavior of any to cover ever greater numbers of objects, the
one of the name relations will be increasingly verbal community combines names in this
evoked by the other name relations and vice manner to specify listener behavior more pre-
versa. Thereafter, each name that has been cisely (see Vygotsky, 1934/1987, p. 276).
intraverbally linked, when it occurs alone may
now contain some of the listener behavior of Naming and Manding
the others. This effect is illustrated in the ex- Thus far the present account has not ex-
ample of a child who, after having touched plicitly dealt with manding, a class of verbal
hot objects (e.g., a radiator and a cup of tea), behavior that features prominentiy in Skin-
is taught to say "hot" by her caregiver. When ner's account. Skinner (1957) describes the
the child sees these objects, she says "hot," mand relation as "a verbal operant in which
which in turn occasions avoidance behavior the response is reinforced by a characteristic
and elicits other emotional reactions. If the consequence and is therefore under the func-
child has previously learned the name "ket- tional control of relevant conditions of de-
tie," the caregiver saying "hot kettle" can ex- privation or aversive stimulation" (pp. 35-
tend the child's listener behavior to the word 36). For example, a hungry infant may learn
hot to this other potentially hazardous item. to say "milk" when this vocal behavior is re-
The child, when she next sees a kettle, re- inforced by her caregiver giving her milk.
peats "hot ketfie," and the listener behavior However, just as was the case for echoic be-
already established to "hot" (avoidance and havior, tacting, and intraverbals, we propose
conditioned emotional responses) occurs at that manding may exist either with or without
the same time as she says the word "kettle," the speaker's own listener behavior to the vo-
which in turn comes to evoke similar listener cal stimulus produced.
behavior. From this point on when the child Within Skinner's broad definition of verbal
sees a ketfie, either with a caregiver present behavior, it is true that some of the child's
or on her own, it may occasion not only "ket- earliest verbal behavior might be classified as
tde" but also, intraverbally, "hot" and its at- mands. For example, the child, when she is
 19383711, 1996, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1901/jeab.1996.65-185 by Capes, Wiley Online Library on [26/05/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
NAMING AAM OTHER SYMBOLIC BEHAVIOR

hungry, points to a specific object such as acquires a tact the corresponding mand ap-
milk, or she cries and makes other noises un- pears collaterally" (p. 16). Skinner (1957, p.
til the caregiver responds by giving her milk 188) gives the example of a child in a toy
(Skinner, 1957, p. 465). However, numerous store who is unable to identify a toy and is
studies have shown that in the course of told that it is "a doodler" (in our terms the
learning to produce speech that approxi- toy has been named doodler). The child im-
mates that of the verbal community, children mediately says, "buy me a doodler." Evidence
generally learn to name objects and events that names are incorporated in name-mand-
before they learn to mand them (e.g., Anis- ing in this manner comes from observational
feld, 1984, p. 92; Halliday, 1975; Lewis, 1936, studies with young children (Huttenlocher &
pp. 158-159; Nelson, 1973; Terrace, 1985; see Smiley, 1987; Moerk, 1992, p. 167) and ex-
also pp. 201-202). When manding comes perimental studies with a 4-year-old boy who
about it does so generally via naming, that is, had language difficulties (Salzinger, Feldman,
names are first established and then are func- Cowan, & Salzinger, 1965) and deaf students
tionally extended to mand objects and events. who used signing (Hall & Sundberg, 1987).
For example, while reaching toward her bot-
tie of milk the thirsty child may utter the Speech for Others and Speech for Self
name "milk," and this may be reinforced by There are two main contexts in which nam-
caregivers giving her some milk and not, as ing occurs. In one the child has as listener
previously, simply saying "yes it is milk" or both herself and one or more others; speech
"good girl" and so on (Skinner, 1957, p. in this context is termed social speech or speech
189). In this way naming rapidly becomes for others. The second context exists when the
multifunctional so that the child is now able only audience entering into the control of
to mand any object or event that she can the behavior is the speaker herself as her own
name. Thus most mand relations (i.e., what listener; this is termed inner speech or speech for
we shall term name-manding) observed in hu- self (Diaz, 1992; Goodman, 1981; Goudena,
man verbal behavior incorporate naming, 1992; Skinner, 1957, pp. 438-452; Sokolov,
with all of the speaker-listener functions and 1972; Vygotsky, 1934/1987, p. 257; Wertsch,
bidirectionality thereby entailed (Huttenloch- 1979a, 1979b; Zivin, 1979). We consider
er & Smiley, 1987). As Skinner (1957) has ob- speech for self here because of its importance
served, "It is possible that all mands which in our later account of such issues as rule-
are reinforced by the production of objects governed behavior, thinking, and equiva-
or other states of affairs may be interpreted lence, particularly because many of the verbal
as manding the behavior of the listener and determinants of these phenomena may occur
tacting [we would say naming] the object or at the covert level out of the easy access of
state of affairs to be produced" (p. 189). experimenters and, perhaps, even of subjects
Some experimental work (Lamarre & Hol- themselves.
land, 1985; Lee, 1981) has suggested that Although it has often been asserted that
manding and tacting are functionally inde- differences in the forms of speech that occur
pendent. However, these were studies con- in these two contexts do not arise until later
ducted with children considerably older than in childhood (i.e., around 3 to 7 years of
those who are the focus of this paper, and age), Furrow (1984a) has shown they are
they employed complex propositions as ver- present in the verbal behavior of children as
bal stimuli (e.g., "where do you want me to young as 2 years old. In his study, Furrow de-
put the object?") and as responses (e.g., "on fined speech for others as any utterance that
the left"). Whatever the relation between was accompanied by the child making eye
what these authors call manding and tacting contact with the caregiver, or that occurred
under such conditions, for present purposes in the preceding or following 2 s of making
it is important to recognize that once naming eye contact. Speech for self was defined as
of an object is established, then, after direct utterances that did not meet the eye-contact
training on only a few mand relations, name- criterion and were not accompanied by other
manding often follows. Thus Lamarre and social interactions with the caregiver. Among
Holland (1985) acknowledge that "in natural several differences observed, it was found
settings it often seems that when a speaker that the child's name-manding of the behav-
 19383711, 1996, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1901/jeab.1996.65-185 by Capes, Wiley Online Library on [26/05/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
212 PAULINEJ HORNE and C. FERGUS LOWE
ior of caregivers occurred more frequently in scribed as verbal thinking. For example, in
speech for others, whereas in speech for self describing the special characteristics of the
there was a higher incidence of self-manding speaker who is his or her own listener, he
(that is name-manding of her own behavior; writes,
see Skinner, 1957, p. 440) and otherwise He speaks the same language or languages
naming her own activities. In a second study and has had the same verbal and nonverbal
(Furrow, 1984b) found that speech for self experience as his listener. He is subject to the
was quieter and of lower and less variable same deprivations and aversive stimulations,
pitch than was speech for others. and these vary from day to day or from mo-
Why these differences exist between speech ment to moment in the same way. As listener
for self and speech for others is an issue that he is ready for his own behavior as speaker at
has not yet been fully resolved, but our ac- just the right time and is optimally prepared
count indicates that the consequences of the to "understand" what he has said. Very little
two types of verbal behavior differ from the time is lost in transmission and the behavior
outset and, as name learning proceeds, are may acquire subtle dimensions. It is not sur-
prising, then, that verbal self-stimulation has
responsible for progressively differentiating been regarded as possessing special properties
the two. When the child first names objects and has even been identified with thinking.
or events, both she and her caregiver listen (1957, pp. 438-439)
to her utterance and respond to it in partic-
ular ways. But when, in the absence of any The transfer of listener behavior between
stimulation from the caregiver, the child names that we have already described for in-
names objects and events it sets in train a very traverbal behavior is greadly enhanced and
different kind of verbal relationship in which further developed in speech for self. With in-
the sole listener of her verbal productions is creases in her naming repertoire, the child
herself. And she is a very special listener. can now emit a stream of names, overtly or
Whereas in social speech the child must coverdy, independently of any further stimu-
speak in a form that is effective for the other lation outside the skin other than the object
listener, in speech for self, there is no need, or event that occasioned the first name. To
for example, to speak loudly, to pronounce this extent the child's behavior becomes less
words fully, or even to emit as many words a function of current environmental stimuli
(see Vygotsky, 1934/1987, pp. 266-280). and more a function of the chain or stream
Thus, for example, in the presence of her of verbal generalizations that is her speech
mother and father, the child may make eye for self. With the development, in addition to
contact with her father and say "daddy push naming, of other types of words and of com-
car," which is followed by her father pushing plex forms of word ordering (syntax), the
the car. But when the child is alone or when child's speech is further transformed, but
caregivers are not interacting with her, she that is another story and one that lies outside
may look at the same car and say quietly the confines of the present paper.
"push ... push." Because she is already look- The Origins of Rule-Governed Behavior
ing at the car, "push" alone is sufficient to and Verbal Control
occasion her own behavior of pushing the
car. As Skinner has observed, speech for self, The present account has implications for
whether overt or covert, shows "very little in- the distinction drawn by Skinner (1969, p.
traverbal chaining ... [in our example, 147) between contingency-shaped and rule-
"push"] ... because it is intimately connect- governed behavior (Catania, in press; Cerutti,
ed with concurrent nonverbal behavior [e.g., 1989; Hineline & Wanchisen, 1989; Zettle &
looking at the car]. The chaining is from ver- Hayes, 1982). Although contingency-shaped
bal to nonverbal and back again ... [and is] behavior is embraced by the three-term con-
... closer to productive verbal thinking for tingency (i.e., discriminative stimulus, re-
this reason" (1957, p. 445). sponse and reinforcer), Skinner's definition
As this quote indicates, Skinner was aware of rule-governed behavior as that which is un-
of many of the unique features of speech for der the control of "prior contingency-speci-
self. He also recognized its central impor- fying stimuli" is more problematic. As a num-
tance in the development of what he de- ber of authors have observed (Hayes &
 19383711, 1996, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1901/jeab.1996.65-185 by Capes, Wiley Online Library on [26/05/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
NAMING AND OTHER SYMBOLIC BEHAVIOR 213
Hayes, 1992; Sidman, 1990, 1992; Stemmer, central importance, however, in drawing a
1992, p. 76), Skinner did not make clear what distinction with contingency-shaped behavior
he meant by "specifying" and how stimuli is not that names specify contingencies (they
come to have such a function. The present may or may not do so), but that they specify
account helps to clarify this problem because the subject's own listener behavior toward
naming is, of course, the principal means by particular objects and events. This is the crit-
which behavior and its consequences are ical feature of verbal control, which suggests
specified. Names are the basic components of that the counter term to contingency shaped
verbal rules and, hence, it is only through an should not be rule governed but verbally con-
analysis of the name relation that the effects trolled.
of rules on behavior can be understood. As
we show in the next part of the paper, the Naming and Symbolic Behavior
absence of such an analysis is a serious flaw We shall next very briefly outline how an
in theories that use the construct of equiva- analysis of naming such as this provides an
lence in an attempt to put right the deficien- approach to dealing with some of the com-
cies in Skinner's account. plex issues in human language research. The
We shall not attempt a detailed analysis of aim is to extend our terminology in ways that
complex rule-governed behavior as it occurs will do justice to the complexity of the phe-
in adults, but will indicate how the earliest nomena and can be shared by others in the
forms of such a relation come about in scientific community, while at the same time
young children. Take, for example, a child maintaining the conceptual rigor of behavior
who from birth has had among her toys a analysis.
plastic bowl. Initially she responds to this Verbal understanding. As we have shown, the
stimulus in ways much like any other non- child begins as a listener to the utterances of
verbal organism would; her seeing it may, for others, but the critical development occurs
example, set the occasion for her chewing it, with the advent of naming when she produc-
handling it, hitting it on the ground, and so es her own utterances and responds to them
on. Having been previously taught the name as she has done to those of her caregivers,
"boat" and shown how toy boats function that is, in a socially shared manner (cf. Mead,
(e.g., how they float, sink, and can be 1934, pp. 89-90, 190-191; Vygotsky, 1934/
pushed along with toy passengers etc.), if she 1987, p. 163). In this sense the child now un-
is now told that the plastic bowl is a boat, derstands her own words. In addition, howev-
her behavior toward the bowl may be trans- er, it is increasingly the case that this listener
formed as she applies to it many if not all of behavior, or understanding, evokes further
the kinds of listener behavior she has ac- speaking. Through continuing social inter-
quired before with toy boats. Her new be- actions, her verbal understanding progres-
havior of floating the bowl in water, pushing sively converges upon the way the verbal com-
it along, and so on may never have been re- munity conventionally categorizes and
inforced in the past but is controlled by the otherwise responds to everyday objects and
name relation (i.e., the verbal stimulus) that events (see Skinner, 1957, p. 278).
the bowl now evokes. (If her caregiver Verbal communication. Because, for each of
named the bowl a "hat," then her behavior the child's name relations, the class of stim-
toward it would be very different.) Behavior ulus events and the listener behavior to which
that, like this, is under the control of name it applies have been largely established by the
relations may be termed verbally controlled be- caregiver, they are also part of the caregiver's
havior (cf. Mead, 1934, p. 108). corresponding name relation. This provides
With the development of naming, particu- the basis for verbal communication between
larly in intraverbal behavior and speech for child and caregiver. Thus, when either the
self, verbal control becomes increasingly child or the caregiver says "pat dog," their
more complex and pervasive. When the child mutual response will be to orient to any one
says, for example, "push boat" or, while look- of the class of objects which they both name
ing at the boat, just "push," such naming en- as "dog" (to the exclusion of other objects)
ters into the control of her own behavior (see and to engage in the further shared behavior
pp. 211-212, Speech for Others ... .). What is of of patting it. Clearly, although these words
 19383711, 1996, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1901/jeab.1996.65-185 by Capes, Wiley Online Library on [26/05/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
214 PAULINEJ. HORNE and C. FERGUS LOWE
may be occasioned by some stimulus events turn, give rise to a range of sensations and
and evoke some listener behavior that they emotional activity that reinforces naming.
do not fully share, it is the stimulus class cat- Thus, when the child names, she not only
egories and listener behavior that they do makes an utterance but also at the same time
have in common that makes verbal commu- brings into being other behavior, either full-
nication possible. It also provides the basis for blown or incipient, overt or covert, all of
caregivers to extend membership of the class- which is bound up with the word.
es embodied within the child's name relation The circular relation between, for example,
and so to develop the child's verbal commu- seeing, saying and hearing, and seeing again
nication. Given that caregivers' orienting and with all of the attendant listener behavior and
other listener behavior are shared with many its consequences is what maintains the flow
others in their verbal community, the child is of listener behavior. As we have shown above,
now also well placed to begin to communi- during each occurrence of naming, listener
cate with the community at large. behavior changes and, in turn, alters the
Verbal meaning. A basic and central feature name relation itself; the name relation is
of naming is that it relates to classes of stimuli therefore not so much circular as a spiral of
(Mead, 1934, p. 88; Vygotsky, 1934/1987, p. development that grows in content with each
249); this, of course, is why the study of nam- occurrence or revolution.
ing has such importance for the literature on Verbal thinking. It may be argued that think-
stimulus equivalence and vice versa. Mead ing is simply behavior and is thus common to
(1934, pp. 84-85) has maintained that when all animal species (see Skinner, 1957, p. 449).
we name a stimulus, we in fact re-cognize it as However, according to the present account
an exemplar of a class and react to it in all of the name relation represents a very special
the ways we have learned, through the cul- kind of behavior, with very special accom-
ture, as appropriate for that class of events; plishments and effects upon the rest of the
in this way we recognize each object as an behavioral system, that may be termed verbal
instance of a universal. Similarly, Skinner thinking. In the final chapter of Verbal Behav-
(1957) has argued that "the verbal response ior Skinner (1957) describes the special prop-
chair is as abstract as red. It is not controlled erties of speaker-listener effects in speech for
by any single stimulus" (p. 110). According self that he also argues should be character-
to Vygotsky (1934/1987, p. 285), it is this
"generalized reflection of reality" that is the ized as verbal thinking. We have attempted to
distinguishing feature of word meaning. In demonstrate how naming is instrumental in
the present account, this is achieved through bringing this about. We have also tried to
the circular relation between the stimulus show how, with the development of intraver-
class, the utterance, and the class of listener bal behavior and speech for self, verbal think-
behavior, these being inextricably linked in ing becomes a complex interaction between
the name relation. That is, when humans be- sometimes barely articulated words and the
come speaker-listeners and name, they mean. multimodal and crossmodal equivalences (cf.
For each naming event, the initiating stim- Bush, 1993) produced by their related listen-
uli may occur in any modality, but the listener er behavior.
behavior that follows may produce stimula- Reference and representation. It may be said
tion involving all the modalities that have pre- that the name re-fers to a class of objects or
viously been engaged, as part of the name events, because in a sense that is close to the
relation, with the class of objects or events derivation of this word, the namer is carried
named. For example, seeing a sign or picture back via listener behavior to what is named
or smelling her caregiver's baking may occa- (i.e., the referent). Similarly, it can be said
sion for the child the spoken word "cake," that the name re-presents, standsfor, substitutes
which in turn may evoke in the child visions for, or specifies the referent, because when
of cakes based upon all of the multifaceted naming occurs it may evoke in the namer and
views of cakes she has previously had; at the other listeners seeing or visualizing of the
same time she may feel the crumbling texture stimulus class to which the referent belongs
as she touches cake and taste the sweetness and other related listener behavior. As we
as she bites into it. Such stimulation may, in have shown, such representation occurs not
 19383711, 1996, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1901/jeab.1996.65-185 by Capes, Wiley Online Library on [26/05/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
NAMING AND OTHER SYMBOLIC BEHAVIOR 215

only when the referent is present but also havior (see Sinha, 1988). Once naming rela-
when it is absent. tions are established in the child's repertoire,
Verbal consciousness. There is also a sense in a major step has been taken in transforming
which it might be said that naming enables what is essentially a biologically based behav-
us to have representations or to have thoughts of ioral system, governed by unilinear three-
objects or events. When the child learns to term contingency relations, to one of much
produce her own verbal stimulus, such stim- greater complexity in which bidirectional, so-
ulation controls the occurrence of her own cially shared relations between stimulus class-
listener behavior, although whether she does es and behavior predominate, and objects
so will, of course, be determined by the con- and events are represented or held in con-
sequences of such behavior. Thus, whether or sciousness in a way that transcends the spatial
not the object itself is present, the child, and temporal constraints that govern their
through naming, can, for example, visualize occurrence in the real world. The name re-
it, hear it, touch it. (This in turn has pro- lation is the behavioral unit upon which this
found effects upon the range of stimuli and transformation is based.
listener behavior that will be related to the
vocalization.) By reengaging the object within
the ongoing name relation she might be said APPLICATION OF THE NAMING
to hold it in consciousness. What she in fact ACCOUNT TO STIMULUS
holds or sustains is, of course, the class of EQUIVALENCE
stimuli produced by her own listener behav- In the previous sections we have described
ior. Clearly, organisms without naming skills how, via naming, the child's behavior that has
cannot have this type of consciousness (see been established with a single stimulus comes
Mead, 1934, p. 95; Skinner, 1969, p. 233). to generalize to stimuli that are physically
Before naming has been established, the very different and how, within particular con-
child's listener behavior is dependent upon texts, she comes to respond to sets of differ-
real-time stimulation from the environment, ent objects and events as interchangeable. It
the effects of which may be fleeting. On the is phenomena of this kind that are the central
other hand, the stream of stimulation pro- focus of the literature on stimulus equiva-
duced by the child's naming is to a large ex- lence. But according to our account, the
tent free of the spatial and temporal con- "emergent" behavior that occurs in these cir-
straints that beset the real-life stimuli that are cumstances, although indeed not directly re-
re-presented. For example, even if not able at inforced or trained, is simply a consequence
a given moment to experience directly a of different stimuli being part of the same
cake, dog, her mother (or even a match-to- name relation. Might it not be the case then
sample stimulus), the child may, nevertheless, that humans' success on match-to-sample
through naming, visualize or imagine them. tests of stimulus equivalence is also directly
Therefore, when the child learns to name, attributable to the ways in which they name
this determines not only what she brings to the stimuli?
consciousness but when she does so. In this
sense speaker-listener effects have "the magic Common Naming and Stimulus Equivalence
we expect to find in a thought process" We will consider the operation of naming
(Skinner, 1957, p. 447). Being freed of real- within the context of the two types of match-
time constraints in this manner, the child can to-sample procedures commonly used to
now re-call or re-member objects, that is, keep- study stimulus equivalence: auditory-visual
ing close to the derivations of these terms, and visual-visual procedures.
use naming to "call back" or "call to mind" Auditory-visual procedures. Many of the ex-
the listener behavior related to those objects perimental studies of stimulus equivalence
(Mead, 1934, pp. 90-100). have used auditory-visual match-to-sample
Symbolic behavior. As we have shown, names procedures, in which each stimulus class in-
can be said to refer to, represent, stand for, cludes a dictated word (e.g., Green, 1990; Sid-
substitute for, specify, and recall classes of ar- man, 1971; Sidman & Cresson, 1973; Sidman,
bitrary but conventionally related objects and Cresson, & Willson-Morris, 1974; Sidman,
events. In this sense, naming is symbolic be- Kirk, & Willson-Morris, 1985; Sidman & Tail-
 19383711, 1996, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1901/jeab.1996.65-185 by Capes, Wiley Online Library on [26/05/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
216 PAULINEJ. HORNE and C. FERGUS LOWE
by, 1982; Sidman, Willson-Morris, & Kirk,
1986). For example, in Sidman's (1971) early
study (see Figures 1 and 2) the sample stimuli
Class 1
were dictated words and the comparison stim-
uli were either pictures or printed words. As
we have already shown, in such conditions
subjects may echo the dictated words and, in
so doing, learn to use the dictated word (e.g., A1 B1 Cl
"car") as their name both for the picture and
the printed word. In support of this analysis,
it should be noted that in the Sidman (1971)
study, although the matching procedure did
not require subjects to name the stimuli, all Class 2
emitted the "correct" class names when pre-
sented with the pictures and the printed
words. Clearly, although it might not be the
experimenter's intention, these procedures
are often effective means of establishing com-
mon names and, hence, for achieving success
on equivalence tests (see pp. 206-207). Al- A2 B2 C2 D2
though since Sidman's (1971) experiment Fig. 15. An example of stimulus sets used in studies
several studies have investigated equivalence of visual-visual match to sample where Al, Bi, Cl, Dl
classes using auditory-visual match-to-sample and A2, B2, C2, D2 are the experimenter-defined stim-
ulus Classes 1 and 2, respectively (after Saunders &
procedures, often using complex stimulus ar- Green, 1992, p. 229).
rangements, the basic account of the role of
naming illustrated in Figure 4 applies in prin-
ciple to all of this research. nevertheless occasion naming. That is, faced
Naming tests in auditory-visual procedures. It with novel "abstract" stimuli, the subject may
is, of course, difficult when, as in these pro- search for features to which she can give a
cedures, the experimenter provides the right common name (see Wulfert, Dougher, &
conditions for common naming to be estab- Greenway, 1991). In the Class 1 stimuli of Fig-
lished, to argue that naming does not have a ure 15, for example, some subjects may "see"
critical role to play in subjects' success on the the Cl stimulus as resembling the letter R,
tests for equivalence. Yet this is what a num- which may occasion the verbal response "R"
ber of investigators contend. Their evidence and related listener behavior; this may lead
is based almost exclusively on the finding that the subject to look for an R in other stimuli,
in postexperimental naming tests presented particularly those that, when paired with Cl
to subjects after completion of the match-to- (as sample and comparison in the matching
sample procedures, some individuals do not task), produce reinforcement. Thus one can
reliably produce common names for the stim- "see" an approximation to R in the Dl and
uli in the same class (Green, 1990; Sidman et Bi stimuli and, if rotated, a lower case r in
al., 1985, 1986; Sidman & Tailby, 1982). all four Class 1 stimuli. Similarly one can
There are, however, major problems with "see" a C or loop feature in the Class 2 stim-
conclusions based on such evidence; these uli. Clearly, the identification of a common
are considered below. feature in these stimuli is determined notjust
Visual-visual procedures. Other studies have by their formal characteristics but by means
used visual-visual procedures, which one of the class of listener behavior evoked dur-
might think would avoid some of the diffi- ing naming. For example, the name "R"
culties that are inherent in auditory-visual evokes listener behavior not only to upper
matching, especially if the stimuli used are case but also to lower case exemplars and can
abstract forms for which there are no existing thus embrace figures that show some resem-
conventional names (see, e.g., Figure 15, blance to both. The subject's naming of the
from Saunders & Green, 1992). In fact, how- stimuli thus transforms the task from arbi-
ever, stimulus arrays, or parts thereof, may trary to nonarbitrary match to sample.
 19383711, 1996, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1901/jeab.1996.65-185 by Capes, Wiley Online Library on [26/05/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
NAMING AND OTHER SYMBOLIC BEHAVIOR 217
Of course, just as identifying a nameable dures should generally take less time to learn
feature that can then be seen in other mem- the baseline relations and be more successful
bers of the class can help a subject to perform on equivalence tests than subjects on visual-
successfully on a matching task, so too a visual procedures. This result is just what is
name that does not conform to class bound- found in match-to-sample studies that have
aries may actively impede successful perfor- compared the two procedures (Green, 1990;
mance. For example, in a study of equiva- Lipkens, Hayes, & Hayes, 1993; Sidman et al.,
lence formation in six 4- to 5-year-old 1986).
children, Dugdale and Lowe (1990) found Naming tests in equivalence procedures. When
that when a common name was given for subjects in studies of visual-visual match-to-
stimuli in each of two classes (named "omni" sample have been presented with naming
and "delta," respectively), for most subjects tests, they may have names for individual
this facilitated both their learning of the base- stimuli, but frequently they do not report a
line relations and, subsequently, their passing common name for each stimulus class; this
of equivalence tests. But for some subjects observation has been interpreted as provid-
emergence of successful performance was de- ing further support for the view that naming
layed by their already having well-established is not necessary for the formation of stimulus
names for some of the stimuli (e.g., "green" equivalence (Green, 1990; Lazar, Davis-Lang,
and "red" for green and red stimuli, respec- & Sanchez, 1984; Sidman et al., 1986). How-
tively). In such cases existing individual stim- ever, the central difficulty that besets almost
ulus names may prevent or delay the acqui- all studies, both auditory-visual and visual-vi-
sition of the new experimenter-given sual, that purport to show that naming is not
common name. It follows that these "inter- involved in successful performance on equiv-
fering" names must be attenuated or alter- alence tasks is that they have not attempted
native verbal behavior devised in order to en- to record subjects' naming behavior in the
sure success on the equivalence task course of match-to-sample experiments but
(Dugdale & Lowe, 1990; cf. Bentall, Dickins, instead have relied upon postexperimental
& Fox, 1993). It should also be noted, how- naming tests in which the subject is usually
ever, that because reinforcers are normally provided with one of the stimuli and is asked,
only contingent upon correct sample-com- for example, "What is it?" (Sidman et al.,
parison selections during baseline match-to- 1985) or "What is the name for this?"
sample training, of the many names occa- (Green, 1990). The names evoked under
sioned by stimulus features, only those that these conditions, however, may not corre-
evoke listener behavior congruent with the spond with the verbal responses of the subject
stimulus class as specified by the experiment- during the experiment itself. This is shown in
er will lead to responding that will be rein- studies conducted by Lowe and colleagues
forced. (see Dugdale & Lowe, 1990): In two separate
Whether or not naming is established in experiments, one with mentally handicapped
ways that facilitate the passing of equivalence adults (Hird & Lowe, 19854; see also Hird,
tests is dependent on the particular training 1989) and the other with children aged 2 to
procedure used. Insofar as auditory-visual 5 years old (Lowe & Beasty, 1987; see also
procedures have many of the characteristics Beasty, 1987), all verbalizations made by sub-
of everyday naming (i.e., a word is spoken by jects during training and testing on match-to-
the experimenter immediately prior to each sample tasks were recorded. Although for
of many presentations of a stimulus), then some individuals there was close correspon-
common naming should be established more dence between the names they produced on
easily with this method than with visual-visual a postexperimental naming test and those
procedures. In the latter procedure, rather they uttered spontaneously in the course of
than the experimenter providing the names the experiment, for others the experimen-
at the outset, names are produced, over the
course of the experiment and through inter- 4Hird, J., & Lowe, C. F. (1985, April). The role of lan-
guage in the emergence of equivalence relations I: Evidence from
action with the reinforcement contingencies, studies with mentally handicapped people. Paper presented at
by the subjects themselves. According to this the annual conference of the Experimental Analysis of
account, subjects on auditory-visual proce- Behaviour Group, York, United Kingdom.
 19383711, 1996, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1901/jeab.1996.65-185 by Capes, Wiley Online Library on [26/05/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
218 PAULINEJ. HORNE and C. FERGUS LOWE
ter's questions prompted very different ple to a green, rather than a red, comparison
names or even analytical descriptions of the (AlBI relation) and a horizontal line sample
stimuli of a kind not previously vocalized in to the red, rather than the green, comparison
the course of the experiment. The weakness stimulus (A2B2). In the second training
of naming tests as evidence has also been phase they were taught to match a vertical
highlighted by Stoddard and Mcllvane line sample to a triangle, rather than cross,
(1986): comparison stimulus (AlCI) and a horizontal
Do these data [from naming tests] lay to rest line sample to the cross (A2C2). Equivalence
the question of response mediation as the crit- tests were then presented to assess whether
ical basis for stimulus equivalence? Probably the subjects could match green to triangle
not.... Some examples may serve to illustrate (BICI) and triangle to green (CIBI) and
the difficulty of this research question. Sup- match red to cross (B2C2) and cross to red
pose a given subject characterizes all the stim- (C2B2). Tests of symmetry were also per-
uli in the entirely visual classes with a common formed to determine whether they could also
descriptive adjectival term, like "rounded," match green to vertical line (Bi to Al) and
"pointed" or "pointing that way," perhaps de- red to horizontal line (B2 to A2) and match
rived from primary stimulus generalization ... triangle to vertical line (Cl to Al) and cross
Alternately, suppose a common descriptive
term, such as "Set 1" vs. "Set 2," was applied, to horizontal line (C2 to A2).
as we do in talking about stimuli within classes. Of the 29 subjects of differing ages who
When asked the question, "What is it?" in re- participated, 17 passed the tests of equiva-
lation to a given stimulus, perhaps the sub- lence following training on the AB and AC
ject's verbal conditioning history had not pre- relations, success being related to age (see
pared him or her to use descriptive terms as pp. 224, 225). Recordings of subjects' spon-
labels, leading to "I don't know" [its name] taneous verbal behavior showed the follow-
responses on the naming test. Would other ing: (a) During the training sessions, all sub-
methods of testing have evoked the descrip- jects named individual stimuli as, for
tions? (p. 157) example, "up," "down," "cross," "triangle, "
But given that researchers record subjects' "green," and "red" (for the vertical line, hor-
verbal reports and find evidence that some izontal line, cross, triangle, green and red
subjects do not use a common name or com- stimuli, respectively). (b) All of those chil-
mon description for all members of a stimu- dren who were successful in the equivalence
lus class, would thisjustify the conclusion that tests had previously, during training, intra-
naming is not necessary for the formation of verbally named the correct sample-compari-
equivalence classes? Some (Green, 1990; La- son pairs: On AlBI trials, for instance, some
zar et al., 1984; Sidman et al., 1985, 1986; but of the children, when presented with the ver-
see also Sidman, 1994, pp. 113, 305-307) tical stimulus as sample and the green stim-
have assumed this to be the case, but there ulus as one of the comparisons, responded
are many other ways, apart from common "up green," and on AlCI trials they said, in
naming, in which names might serve to bring the presence of the vertical and triangle, "up
about success on equivalence tests (see pp. triangle"; similarly, when presented with the
221-222, Other Verbal Behavior ...). One such horizontal stimulus, they said "down red" or
is intraverbal naming. "down cross" (see Figure 16). (c) Some sub-
jects, when presented during baseline ses-
Intraverbal Naming and Stimulus sions with one of the stimuli in a particular
Equivalence class, responded by naming all three class
How intraverbal naming can enable chil- members despite the fact that only two of
dren to pass tests of equivalence is shown in these were present at the time. This also oc-
the Lowe and Beasty (1987) study (see Beasty, curred during baseline and test trials in
1987; Dugdale & Lowe, 1990), which record- equivalence test sessions; for example, when
ed the spontaneous verbalizations of children presented with the vertical line (Al), 1 child
aged 2 to 5 years old while they performed responded "up green, up hat" (where her
on a visual-visual match-to-sample task (see name for the vertical Al stimulus was "up,"
Figure 16). In the first training phase the chil- for the Bi green stimulus "green," and for
dren were taught to match a vertical line sam- the Cl triangle stimulus "hat"); another
 19383711, 1996, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1901/jeab.1996.65-185 by Capes, Wiley Online Library on [26/05/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
NAMING AMD OTHER SYMBOLIC BEHAVIOR 219

B1 B2
AB

II
I I
BC ICB A 2
BC *CB
A1[ I I

'I

I.
AC AC

C1 C2

|Key: * Trained R = Red

m'-m m
_ Tested G = Green
Fig. 16. The equivalence paradigm employed by Lowe and Beasty (1987). The arrows point from samples to
corresponding comparison stimuli. Solid arrows indicate trained relations (AB and AC), and broken arrows depict
relations assessed during testing (BA, CA, BC, and CB). The experimenter-defined stimulus Classes 1 and 2 are shown
in the left and right displays, respectively.

child, when presented with the red stimulus If she then selects the red stimulus in re-
(B2), said "down red cross," naming A2, B2, sponse to her last saying "red," no reinforcer
and C2, respectively. will be forthcoming and the probability of
How intraverbal naming of the correct saying "red" and selecting red in the pres-
sample-comparison stimulus pairs could give ence of the vertical stimulus will decline. If,
rise to success on the equivalence tests follows however, on this or a subsequent trial with the
from our general account. In the course of vertical sample she says "green" and then se-
training on the match-to-sample procedure lects the green key, this behavior will be re-
employed in the Lowe and Beasty (1987) inforced and the probability of saying
study and most others on equivalence, the "green" and selecting the green key in the
subject is likely to name the stimuli, either presence of the vertical will increase. As the
overfly or coverdy, when she sees them. Thus, child learns the AB relations, she will thus in-
as shown in Figure 17, when presented with creasingly tend to name successful sequences
the vertical stimulus as sample the child, like of responding to stimuli (e.g., "up green" or
some of the subjects in the Lowe and Beasty "down red"). (This is a process similar to that
study, may name it "up" and then press that discussed on pp. 209-210, How intraverbals
response panel. When the green and red ... . and by Skinner, 1957, pp. 75-76, where
comparison stimuli then appear, she may ini- the child comes to say "spoon fork" when
tially name them both by saying "green red." regularly seeing both together.) In this case
 19383711, 1996, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1901/jeab.1996.65-185 by Capes, Wiley Online Library on [26/05/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
220 PAULINEJ HORNE and C. FERGUS LOWE

Key: G = Green
R = Red
Fig. 17. Subjects' intraverbal naming of the stimuli in the Lowe and Beasty (1987) study. In learning baseline
relations (AB and AC) the subject simultaneously learns bidirectional intraverbal name relations "up" + "green,"
"up" <-+ "triangle," "triangle" +-+ "green" (left) and "down" *4 "red," "down" *-> "cross," "red" e-* "cross" (right)
between the stimuli (gray arrows). When the subject names any stimulus presented as sample, intraverbal naming of
its other two class members is evoked; this, in turn, occasions selection of whichever of the two is presented as the
comparison in that particular trial. Thus, through intraverbal naming, stimuli within a class become interchangeable
in the context of match to sample.

the child continues to name the correct sam- obtain when testing for symmetry between
ple-comparison stimulus pairs because this B2-A2 (i.e., red and horizontal stimuli). (It is
behavior controls correct responding on the also, of course, possible at this stage that oth-
match-to-sample task (i.e., enables the child er verbal behavior, e.g., "green goes with up"
to "remember" what goes with what) and is or "up goes with green," may be involved; see
thus reinforced through the experimental below.)
contingencies. As was frequenfly observed in Intraverbally produced bidirectional rela-
the Lowe and Beasty study, the child may self- tions between A and C stimuli may emerge in
echoically repeat "up green up green up a fashion similar to those for A and B. At the
green," which establishes a bidirectional re- start of training on AC relations the child,
lation not only between the verbal responses when presented with the A stimulus (e.g., the
"up" and "green," but also between their re- vertical), may continue for some time with AB
lated listener behavior of responding on the verbalizations (e.g., "up green") but because
vertical and green keys. So, as shown in Fig- B stimuli (e.g., green) never follow the A sam-
ure 17, upon testing for the symmetrical Bi- ple in these trials and only AC responding is
Al relation, the child, seeing the green stim- reinforced, the child learns when given the A
ulus (BI) as a sample for the first time, says sample (e.g., vertical) to select the C stimuli
"green"; the resultant verbal stimulus /green/ (e.g., triangle) and at the same time to name
occasions intraverbally her saying "up" and intraverbally the correct AC stimulus pairs
hearing /up/ and then in turn selecting the (e.g., "up triangle"). When AB and AC trials
vertical stimulus (Al). Similar relations will are then mixed with equal probability, as is
 19383711, 1996, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1901/jeab.1996.65-185 by Capes, Wiley Online Library on [26/05/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
NAMING AM\) OTHER SYMBOLIC BEHAVIOR 221

standard practice in these procedures, each common naming strategy recorded for some
sample (A) name will intraverbally give rise of the subjects in the Lowe and Beasty (1987)
to two different names of comparison stimuli study. Similarly, the horizontal, red, and cross
(B and C) including their related listener be- stimuli may each occasion "down." As was
havior. Thus, as shown in Figure 17, when the shown earlier (see Figure 14), common nam-
vertical stimulus (Al) is the sample, the child ing brings with it novel behavior, in this case
may say "up green up triangle" or "up tri- selecting the other within-class stimuli when-
angle up green"; because the vertical stimu- ever one member of the class is presented as
lus occasions "green" and "triangle" with the sample, thus meeting all of the test cri-
similar probabilities, /up/ may also give rise teria for stimulus equivalence.
to "green triangle," yielding now in response
to the vertical sample the three-name intra- Other Verbal Behavior and
verbal "up green triangle" (or "up triangle Stimulus Equivalence
green"); intraverbal sequences of just this So far we have shown how two particular
kind were articulated by subjects in the stud- forms of verbal behavior-common naming
ies by Lowe and Beasty (1987) and Hird and and intraverbal naming-can bring about
Lowe (1985). With self-echoic repetition of success on tests of equivalence. But there are
these sequences, any of the A, B and C stim- other possibilities. For example, the young
ulus names may come to occasion the other children in the Lowe and Beasty (1987) study
two. At this point an intraverbal "equiva- produced a variety of descriptions of the stim-
lence" class has been formed, wherein the ulus relations as they performed the task
names that the subject has applied to the ex- (e.g., "green means up," "green is up,"
perimental stimuli in each class become in- "green is the same as the triangle," "what
terchangeable. This leads directly to success does red go with?" "they [pointing to the ver-
on formal tests of equivalence within the tical and triangle] are both green"). Similar
matching paradigm. Figure 17 shows, for ex- findings have been reported by Wulfert et al.
ample, that on CB test trials, when the child (1991) who, in an interesting methodological
for the first time sees the triangle (Cl) as a departure for equivalence research, used a
sample, she names it "triangle," which in- "think aloud" procedure modeled upon Er-
traverbally occasions her saying "up" and icsson and Simon's (1980) protocol analysis.
"green." Having pressed the sample, the The results of their first experiment with
green (Bi) and red (B2) keys appear as com- adult humans showed that subjects who
parison stimuli: Hearing herself say /green/, formed equivalence classes described the re-
she orients to and selects the green, rather lations among the stimuli in each class: They
than the red, key. Similar relations should named and linked the stimuli with phrases
hold between BI-Ci, C2-B2, and B2-C2 such as "circle goes with the open triangle,"
names and respective stimulus selection. which also occurred in more telegraphic in-
Such behavior is taken to indicate that sub- traverbal form (e.g., "circle triangle"), as in
jects have formed stimulus equivalence. the Lowe and Beasty study. In addition, on
From intraverbal to common naming. There is, equivalence test trials the successful subjects
however, another verbal route that may de- explicitly named the nodal stimulus (i.e., the
velop in the course of learning these condi- sample stimulus that during training was
tional relations. For example, if on Al-Bi and common to all of the stimulus pairings within
Al-Cl trials a subject names the stimulus each class). As the study progressed, 2 of the
pairs "up-green" and "up-triangle," respec- subjects ceased to utter the names for the in-
tively, then because "up" (the name of the dividual stimulus relations in each class and
nodal stimulus) is common, both the intra- instead used a common "category" name.
verbal name pairs may contract to the com- Those subjects who did not show equivalence
mon name "up." This is particularly likely to did not describe the relation between stimuli
happen with repeated presentations of Al-BI within each experimenter-defined class but
and Al-Cl trials. Thus, as depicted in Figure instead verbally linked together stimuli from
18, whenever not only the vertical stimulus different classes on the grounds that they be-
but also the green or the triangle stimulus are longed together as a compound (e.g., "To-
presented all three may be named "up," a gether they look like a person with a hat") or
 19383711, 1996, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1901/jeab.1996.65-185 by Capes, Wiley Online Library on [26/05/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
222 PAULINEJ. HORNE and C. FERGUS LOWE

66 up99 /Up/
up Iiil "down"' Idownl

B1
I,

,1 1
,, A

I
, .1
I C1
I L,
1L _

Key: Trained G = Green

----- Untrained R = Red

Fig. 18. When a subject comes to emit the common name "up" in relation to each Class 1 stimulus (left) and
"down" in relation to each Class 2 stimulus (right), new relations between stimulus class members result (broken
arrows), thereby fulfilling the criteria for stimulus equivalence.

referred to across-class common physical fea- psychologists' criteria for stimulus equiva-
tures of the stimuli, thus cutting across the lence. If the supposed absence of common
stimulus class groupings as specified by the naming of stimuli within the experimenter-
experimenter. This study, like that of Lowe defined classes is the only developed argu-
and Beasty, shows how both intraverbal nam- ment against the naming hypothesis, it must
ing of the stimulus-comparison pairs and be concluded that the latter is far from being
common naming can determine perfor- disproved.
mance on these tests.
Naming tests, naming, and stimulus equiva- Key Tests of the Naming Hypothesis
lence. It follows from the foregoing that if ev- A major virtue of the naming hypothesis of
idence were to be found that subjects in any stimulus equivalence is that it gives rise to sev-
given equivalence experiment did not have eral clear predictions that are open to exper-
common names for stimuli in the same class, imental disconfirmation. Fortunately, a num-
that would not, in itself, justify the conclusion ber of them have already been submitted to
that naming in some form is not necessary experimental test:
for success on these tasks. What the present Because they are lacking in naming skills, non-
analysis shows is that once language is estab- human organisms will generally fail tests of stim-
lished there may be, in addition to common ulus equivalence. There is no theoretical basis
naming, a large, possibly infinite, number of within current accounts of nonverbal animal
ways in which naming can enter into subjects' learning for assuming that stimulus equiva-
choices of stimuli such that they conform to lence, at least as conventionally assessed,
 19383711, 1996, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1901/jeab.1996.65-185 by Capes, Wiley Online Library on [26/05/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
NAMING AND OTHER SYMBOLIC BEHAVIOR 223

should ever arise (Catania, 1992, p. 156). For 1987; Vaughan, 1988), others have argued
example, the three-term contingency, which that their data do not meet the necessary cri-
is central to Skinner's (1938) account of op- teria for equivalence (see Dugdale & Lowe,
erant conditioning, specifies a relation link- 1990; Hayes, 1989; Saunders, 1989).
ing a discriminative stimulus, a response, and Recently, Schusterman and Kastak (1993),
a reinforcer, such that in the presence of the using a novel match-to-sample procedure, re-
stimulus the response is reinforced. The op- ported that a sea lion succeeded in passing
erant literature shows that such discrimina- tests of stimulus equivalence. Having first de-
tive functions may take many forms and be vised 30 potential three-member stimulus
of considerable complexity, but characteristi- classes they designated 12 of these for train-
cally the direction of the relation between the ing of symmetric and transitive relations and
discriminative stimulus and response is one reserved 18 classes for equivalence tests. They
way. Thus if standard match-to-sample pro- trained AB and BC relations with all 30 class-
cedures are used to establish contingencies es and used the following procedure to train
with an animal subject (e.g., a pigeon) such symmetry and transitivity on the selected 12:
that in the presence of a green key (the sam- (a) AB training followed by BA ("symmetry")
ple) a response on another key, a vertical line testing and training, (b) BC training followed
(correct comparison stimulus), is reinforced, by CB ("symmetry") testing and training, (c)
then when the green stimulus is presented we AC ("transitivity") testing and training, and
would expect the probability of responding (d) CA ("symmetry") testing and training.
on the vertical stimulus to increase. There are The sea lion was then tested for CA (equiva-
no theoretical grounds, however, to expect lence) on the 18 reserved stimulus classes,
that when we present the vertical as sample correct responding being reinforced
the animal will respond with high probability throughout. Taking the first test trial for each
on the green comparison key. Indeed, it may stimulus class, performance was correct on 16
be the very effectiveness of the three-term out of the 18 trials, which was significantly
contingency that in part precludes animal better than chance.
species, other than humans, from success on There are a number of distinctive features
equivalence tests; if organisms are locked into of this study that might throw light upon the
such unidirectional relations there is no op- sea lion's success where other animal species
portunity or mechanism by which reversals have failed. First, on all trials, the sample and
can occur. On the other hand, most humans comparison stimuli were simultaneously visi-
are, through naming, freed of the temporal ble to the sea lion before it was released to
and spatial constraints that govern discrimi- respond to one of the comparison locations.
native control of animal behavior because Second, no sample response was required,
they produce their own stimuli (e.g., names) the only response requirement being orien-
that form reversible and other links between tation to the left or right of release location
their behavior and stimuli in the environ- such that the sea lion's nose broke the plane
ment. formed by the front of the correct compari-
Numerous experimental studies support son stimulus. Third, the incorrect compari-
this view and have shown that nonhuman an- son stimuli were randomly assigned and so
imals may learn conditional discriminations differed from trial to trial, minimizing any
readily, but without this resulting in stimulus build-up of CS- or negative discriminative
equivalence. Despite concerted efforts, stud- stimulus control by particular comparison
ies to date have failed to demonstrate equiv- stimuli. Fourth, training was given on a subset
alence with pigeons, cebus monkeys, rhesus of "symmetry" and "transitivity" relations;
monkeys, baboons, and chimpanzees (see correct responding, including that on test tri-
D'Amato, Salmon, Loukas, & Tomie, 1985; als, was reinforced throughout the study.
Dugdale & Lowe, 1990; Hogan & Zentall, Whether these or other features might
1977; Holmes, 1979; Kendall, 1983; Lipkens, have brought about success on CA tests is dif-
Kop, & Matthijs, 1988; Rodewald, 1974; Sid- ficult to assess on the basis of the limited pro-
man et al., 1982). Although some investiga- cedural information available, but one possi-
tors have claimed to find equivalence with an- bility is that in combination they provided the
imal subjects (McIntire, Cleary, & Thompson, conditions to establish each stimulus pair
 19383711, 1996, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1901/jeab.1996.65-185 by Capes, Wiley Online Library on [26/05/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
224 PAULINEJ. HORNE and C. FERGUS LOWE
(e.g., within the AB and BC classes) as a com- have not yet learned to name or in whom
pound stimulus that evoked (a) swaying or naming skills are not yet well established. Un-
moving back and forth (which is food-related fortunately, not very much research on stim-
behavior in sea lions) across the two stimulus ulus equivalence has been conducted with
elements of the compound until the plane in very young children, but studies that do exist
front of the outer element was broken and point to a close correlation between the de-
the fish reinforcer delivered, and (b) as train- velopment of language and success on equiv-
ing progressed, the response of moving to the alence tasks. For example, a study by Devany
outer stimulus element (comparison) of the et al. (1986; but see also Augustson & Dough-
compound. During "symmetry" (BA and CB) er, 1992) tested for stimulus equivalence in
training, with continuous reinforcement for three groups of subjects: normal 2-year-old
correct responding, these effects of stimulus children, 2- to 4-year-old mentally handi-
pairs operating as compounds should be capped children with functional spontaneous
strengthened. On the basis of studies that speech and signing, and 2- to 4-year-old men-
have demonstrated "associative transitivity" tally handicapped children with no functional
in match to sample (D'Amato et al. 1985; verbal skills. Both of the first two groups (i.e.,
Steirn, Jackson-Smith, & Zentall, 1991; see normal and mentally handicapped with lan-
Zentall & Urcuioli, 1993), such that when AB guage) passed the equivalence tests, whereas
and BC are trained, the AC relation is present the mentally handicapped subjects who had
upon testing, it might be expected that the no language failed. In a study of normal and
sea lion would pass the test of AC transitivity; hearing-impaired children, Barnes, Mc-
this was indeed found. But given that the AC Cullogh, and Keenan (1990) have reported a
stimulus elements also come to function as a similar relation between verbal ability and
compound and that, whether presented as success on tests of equivalence.
AC or CA, the animal moves towards the out- Because language development over the
er element of that compound (as described first few years of life in young children is cor-
above), this may be sufficient to account for related with chronological age, we would ex-
successful performance on the continually re- pect, according to the present account, to ob-
inforced CA (equivalence) test trials. serve more failure on equivalence tests by
This account is, of course, highly specula- children in the younger age ranges and more
tive and needs to be submitted to experimen- success by those who are older and have ac-
tal test. Indeed, the study itself remains to be quired better naming skills. This was investi-
replicated. Until then it may not be possible gated in the study by Lowe and Beasty (1987;
to determine whether Schusterman and Kas- see above), whose subjects were in three
tak's procedure constitutes a demonstration different age groups: 4 to 5, 3 to 4, and 2 to
of learned higher order generalized equiva- 3 years old. Following standard training on
lence (Boelens, 1994) or whether simpler the baseline conditional discriminations, (a)
learning principles such as those outlined all 10 of the 4- to 5-year-old children passed
above apply. It should also be noted, however, tests of equivalence, (b) half of the 12 sub-
that if a particular nonhuman animal in a jects aged 3 to 4 years old succeeded, and (c)
particular match-to-sample context ever did only 1 out of 7 of the 2- to 3-year-old subjects
succeed in passing tests of stimulus equiva- passed (see next section).
lence, it would not mean that the functional Teaching subjects particular name relations for
determinants of the behavior would be the the stimuli used in match-to-sample procedures may
same as those which enable verbally compe- be a powerful determinant of subsequent perfor-
tent humans to succeed on these tasks. To mance on equivalence tests. Dugdale and Lowe
assume the determinants were the same (1990) have reported that when normal 4- to
would be an example of a formalistic fallacy 5-year-old children who fail tests of equiva-
(Skinner, 1969). This and related issues are lence are subsequently taught a common
discussed on page 233. name (e.g., "omni" or "delta") for the stim-
Humans who are lacking the prerequisite nam- uli in each experimenter-defined class, they
ing skills will generally fail tests of stimulus equiv- then go on to pass the tests. A common nam-
alence. One of the most fruitful ways to test ing intervention has also been used by Eike-
this prediction is with young children who seth and Smith (1992) in their investigation
 19383711, 1996, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1901/jeab.1996.65-185 by Capes, Wiley Online Library on [26/05/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
NAMING AND OTHER SYMBOLIC BEHAVIOR 225
of visual-visual equivalence in 4 autistic chil- subjects in the nonnaming group, who had
dren. The subjects' ages ranged from 3.5 to initially failed the equivalence tests, were sub-
5.5 years but all showed evidence of deficits sequently instructed to name the stimuli, they
in language skills compared to normal chil- also went on to pass the tests. Saunders et al.
dren. Having learned the baseline condition- recognize that naming instructions have an
al discriminations, all 4 children initially important role in bringing about these effects
failed the tests of equivalence; when common and suggest that the comparison-as-node pro-
naming of the stimuli in each class was then cedure results in "anticipatory naming of the
introduced, 2 of the subjects passed and a 3rd correct comparison for each sample. Thus,
performed at well above chance levels. The every sample within a class would control a
4th subject, Rory, was the only one to per- common naming response" (p. 732).
form at chance levels on the tests and also The question, however, is how this com-
had the lowest level of language development mon naming has its effects: Saunders et al.
in the group, as indicated by both expressive (1993; see also Eikeseth & Smith, 1992) sug-
and receptive measures. In another phase of gest that when different stimuli evoke the
the experiment and outside the match-to- same name, this brings about a functional
sample format, the same 4 subjects were stimulus class that then somehow results in
taught names for entirely new stimuli. On equivalence class formation. However, as we
subsequent match-to-sample tests, 3 subjects have already shown in the first section of this
demonstrated emergent relations between paper, to establish a common response to dif-
the stimuli that had been assigned a common ferent stimuli and thus a functional class is
name, thus meeting the criteria that would not sufficient to yield stimulus equivalence.
normally be set for equivalence with match- To provide an explanation of the latter, in-
to-sample procedures, except that in this case cluding the findings of the study by Saunders
none of the baseline conditional relations et al., requires an account of naming not sim-
had been trained with these stimuli prior to ply as speaking but as speaker-listener behav-
testing. In this phase, Rory was again the one ior, with all of the bidirectional effects there-
with whom the procedure was least success- by entailed.
ful; however, when the procedure for teach- Evidence that interventions based upon in-
ing him names was modified, emergent rela- traverbal naming can be very effective deter-
tions were also facilitated in this subject. This minants of outcomes on equivalence tests
second phase is a good experimental dem- comes from the Lowe and Beasty (1987)
onstration of a phenomenon that occurs dur- study in which the 12 children who initially
ing normal language development in chil- failed the equivalence tests were subsequently
dren (see pp. 205-207, Naming and emergent taught to name the sample-comparison stim-
behavior), and the study as a whole shows that ulus pairs. On AlBI baseline trials (see p.
common naming can be a powerful interven- 218) they were taught to say "up-green," on
tion in bringing about equivalence even in AlCl trials "up triangle," on A2B2 "down
autistic children. red," and on A2C2 "down cross." Teaching
Saunders, Saunders, Williams, and Sprad- this intraverbal naming, which was similar to
lin (1993) presented evidence showing that, that spontaneously produced by the children
for subjects with mild retardation, match-to- who had passed the equivalence tests, led to
sample training involving four pairs of sample 11 of the 12 children passing the BC and CB
and one pair of comparison stimuli (compar- tests. The one 2-year-old child who failed the
ison as node) was much more successful in tests was also the only child who failed to
yielding stimulus equivalence than proce- learn the sample-comparison names in the
dures with four pairs of comparison stimuli time available.
and one pair of samples (sample as node). As well as leading to success, some verbal
Success on the comparison-as-node proce- interventions may also lead to failure on
dure, however, was shown to rely heavily upon equivalence tests. In a study of match-to-sam-
the subjects being given verbal instructions ple performance in normal adults Dickins,
and names for the stimuli. Subjects who were Bentall, and Smith (1993) first taught base-
not provided with names had a much higher line relations, after which subjects in one
failure rate than those given names; when group were taught intraverbal names that
 19383711, 1996, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1901/jeab.1996.65-185 by Capes, Wiley Online Library on [26/05/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
226 PAULINEJ. HORNE and C. FERGUS LOWE
linked stimuli across rather than within ex- sure success, it is not sufficient for subjects
perimenter-defined classes and subjects in a simply to have naming skills (they might all
second group were taught intraverbal se- have these) but more specifically, their stim-
quences unrelated to on-task stimuli. The ulus class naming must be congruent with the
group of subjects taught across-class intra- experimenter-defined classes. The present ac-
verbal sequences were much less successful count of match-to-sample performance not
on subsequent tests of stimulus equivalence only makes sense of the data from equiva-
than were control-group subjects. This study lence studies but is also consistent with a wid-
shows that intraverbal naming can work for er literature that shows many differences be-
or against success on tests of stimulus equiv- tween behavioral relations in humans and
alence depending on whether or not the in- nonhumans that are attributable to the role
traverbal sequences that are formed before of language (Bentall, Lowe, & Beasty, 1985;
such testing are congruent with experiment- Catania, Matthews, & Shimoff, 1982; Hayes &
er-defined classes. Hayes, 1992; Horne & Lowe, 1993; Le Fran-
In a recent paper, Mandell and Sheen cois, Chase, &Joyce, 1988; Lowe, 1979, 1983,
(1994) set out explicitly to test the naming 1989).
hypothesis, reasoning that if naming is an im- It has been suggested (Hayes, 1994; Hayes
portant determinant of stimulus class forma- & Hayes, 1992; Sidman, 1994) that using lan-
tion, then whether or not the stimuli in guage or naming to explain findings from
match-to-sample procedures are easily match-to-sample studies is reminiscent of me-
nameable should be predictive of perfor- diational models prominent in earlier
mance on equivalence tests; the more name- decades. Mediated generalization theory
able the stimuli, the more likely it should be (Cofer & Foley, 1942) was used to deal with
that subjects would be successful. In a visual- a set of behavioral effects that were also
visual match-to-sample study with normal termed equivalence, although they did not
adults, subjects saw three kinds of sample conform exactly with the definition of that
stimuli that varied in pronounceability: (a) term put forward by Sidman, and were stud-
phonologically correct pseudowords (e.g., ied by a different methodology, usually
SNAMB), (b) phonologically incorrect words paired-associate learning (Jenkins, 1963; Jen-
(e.g., NSJBN), or (c) punctuation marks kins & Palermo, 1964; Sidman, 1994). This
(e.g., +]*' !I). Subjects in the group that was was an associative account that used the par-
exposed to the pronounceable stimuli dem- adigm of Pavlovian conditioning, particularly
onstrated equivalence class formation more higher order conditioning, to make infer-
quickly and with fewer errors than did other ences about Pavlovian conditioned responses
subjects. Also, some of the subjects who were that supposedly mediated stimulus presenta-
not provided with sample stimuli that were tions, enabling symmetry, transitivity, and
easily pronounceable nevertheless found oth- equivalence to emerge. It was essentially a
er more idiosyncratic means of naming the stimulus-response chaining model, and like
stimuli (e.g., naming HCKTR as "HECTOR" any such account it had great difficulty in
or naming just the first letter of each non- dealing with the kinds of stimulus-stimulus re-
phonological word). In a second study, sub- versals termed symmetry. Consequently, its ad-
jects were pretrained to apply names to non- vocates had to appeal to backward condition-
phonological words. As a result, their ing (Ekstrand, 1966), normally a very weak
subsequent match-to-sample performance, effect if it exists at all, and chaining to explain
particularly the development of stimulus what were very reliable and robust phenom-
equivalence classes, was considerably en- ena (Jenkins, 1963). Clearly, the present ac-
hanced compared to the performance of sub- count has little in common with traditional
jects without such pretraining. mediated generalization theory; it is not
Overview. The studies reviewed above show based on classical conditioning nor does it in-
that interventions that train common nam- volve appeals to backward conditioning,
ing, intraverbal naming, or other naming be- chaining, or any of the other associative
havior to relate the experimental stimuli are mechanisms of that theory. Indeed the pri-
very effective in determining success on mary role of naming should not be viewed as
equivalence tasks. They also show that, to en- mediating the establishment of stimulus class-
 19383711, 1996, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1901/jeab.1996.65-185 by Capes, Wiley Online Library on [26/05/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
NAMING AND OTHER SYMBOLIC BEHAVIOR 227
es: Naming is stimulus-classifying behavior ulus and response loses significance; indeed,
(see pp. 207-208, 213-215, Verbal communica- as he puts it, "Equivalence relations have
tion, Verbal meaning). But the history of me- their own defining characteristics, none re-
diated generalization is an interesting one quiring the stimulus/response dichotomy"
that has implications for current theories of (1994, p. 386).
equivalence, to which we shall return below. It is important to note that, for Sidman,
In addition to generating falsifiable predic-
tions, any explanation of stimulus equiva- Strictly speaking, reinforcement contingencies
do not create equivalence relations; rather,
lence should be able to account for the ex- they create prerequisites, or the potential, for
isting experimental data. The naming demonstrating the properties that define an
hypothesis succeeds on both these counts, al- equivalence relation. Additional factors, like
though it remains, of course, open to exper- the test conditions, contextual control, and a
imental disconfirmation. We shall next con- subject's behavioral history will help deter-
sider alternative explanations of equivalence mine whether and how that potential is real-
put forward by Sidman and by Hayes and col- ized. ... An equivalence relation, therefore,
leagues.5 In what follows we oudine briefly has no existence as a thing, it is not actually
the main features of these two accounts and established, formed, or created. It does not exist,
subject them to critical but, we hope, con- either in theory or in reality. It is defined by
the emergence of new-and predictable-an-
structive scrutiny, the aim being to further de- alytic units of behavior from previously dem-
bate. We apologize in advance for any errors onstrated units. (1994, pp. 387-388).
or misconceptions in our appraisal; undoubt-
edly, if there are any such, they will be clari- Because responses as well as stimuli enter into
fied in the commentaries to this article. We equivalence relations, functions possessed by
later discuss features that are common to all any one element of a reinforcement contin-
three accounts and suggest ways in which fur- gency giving rise to equivalence will also be
ther experimental work might help to possessed by other elements within it (1994,
achieve greater agreement between them. p. 392). Thus, discriminative stimuli will func-
tion as reinforcers, and reinforcers will func-
Sidman 's Theory of Equivalence tion as discriminative stimuli. Whereas behav-
According to Sidman, equivalence (like re- ior analysts, including Sidman himself, have
inforcement, discrimination, etc.) represents previously viewed conditioned reinforcement
a primitive function not derivable from other as a basic nonderived stimulus function, ac-
behavioral processes (Sidman, 1992, p. 22). cording to this account conditioned rein-
In his recent book (Sidman, 1994), he de- forcement is derived from equivalence rela-
scribes in some detail the conditions that give tions (1994, pp. 391-393). Sidman proposes,
rise to equivalence relations and, departing furthermore, that the prerequisites for an
from his earlier view (e.g., Sidman, 1986) that equivalence relation may also arise during
equivalence is established by four-term con- Pavlovian or respondent conditioning. In-
tingencies, concludes that three-term, or per- deed, he suggests that Pavlovian conditioning
haps even two-term, contingencies could suf- may itself be a derived phenomenon that aris-
fice. Whereas his earlier account was es from the establishment of equivalence re-
concerned with stimulus-stimulus relations, lations between pairs of stimulus events in
he now proposes that as a result of their par- conditioning procedures (1994, pp. 403-
ticipation in reinforcement contingencies, 404). Thus, a stimulus paired with an uncon-
both stimuli (whether discriminative, condi- ditioned stimulus may become a conditioned
tional, or reinforcing) and responses become stimulus by virtue of the participation of both
members of an equivalence relation. Thus, in an equivalence relation.
within the event pairs that define the equiv- To establish which particular stimulus class-
alence relation, the distinction between stim- es will arise from particular reinforcement
contingencies is not, however, a straightfor-
5We are grateful to Murray Sidman and to Steve Hayes ward matter. For example, according to Sid-
for their very helpful and constructive comments on
some of the descriptive material on their accounts pre- man, because responses and reinforcers both
sented here. Any remaining errors of interpretation are enter into equivalence relations, during train-
entirely our responsibility. ing of the conditional relations AlBI and
 19383711, 1996, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1901/jeab.1996.65-185 by Capes, Wiley Online Library on [26/05/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
228 2PAULINEJ. HORNE and C. FERGUS LOWE

A2B2 one large equivalence class (i.e., Sidman also acknowledges that verbal rules
A1B1A2B2) must emerge if a common rein- can establish equivalence relations, although
forcer, a defined response (e.g., key press- rules are not always necessary (1994, p. 509)
ing), or both are employed throughout the and, in any case, he believes that appealing
procedure (Sidman, 1994, p. 408). The prob- to rule following as an explanation begs the
lem then is to account for what appears to question as to where rules come from in the
be, as Sidman himself acknowledges, "an im- first place (Sidman, 1992, pp. 21-22). Ac-
mense gulf between theory and data" (1994, cording to his account, it is the formation of
p. 410), that is, that most studies (at least with equivalence relations that makes rule-gov-
human subjects) repeatedly fail to show this erned behavior possible. He maintains that
merging of classes but show instead "success" when naming or more complex verbaliza-
on experimenter-defined tests of equivalence. tions such as rules become members of an
Sidman's explanation is that the incompati- equivalence class, they become indistinguish-
bility between the reinforcement contingen- able from other stimulus and response events
cies and the formation of an overarching and, as such, do not have a special mediating
equivalence class causes the "selective drop- function. This, he believes, allows one to dis-
ping out" of common elements, that is, the pense entirely with the notion of verbal me-
response and reinforcer from the prerequi- diation of equivalence relations (Sidman,
site class (1994, p. 411). Similarly, given that 1994, pp. 510-511).
the experimental context can enter into In Sidman's view, because equivalence is an
equivalence relations and is common to all of evolutionary given for which verbal skills are
the other experimenter-defined stimulus not necessary, it may occur in a range of an-
classes, it likewise should condense all of the imal species, although its generality remains
classes into one, but in fact does not. Here, to be established (1994, p. 390). Indeed, he
Sidman maintains that because this would suggests that "species differences need not be
wipe out differential control by discriminative absolute but may depend on the ethological
stimuli in each three-term unit and by con- validity of the stimuli and functions being ex-
ditional stimuli in each four-term unit, the amined" (1994, p. 164). Failures to demon-
subject would be unable to meet the terms of strate equivalence with nonhumans may, he
the contingency and because the contingency believes, be due to features of the standard
"calls for differential control" (i.e., what is conditional discrimination procedure em-
termed class intersection in set theory) this ployed in equivalence studies (1994, pp. 166-
takes priority over the formation of the over- 175).
arching stimulus class (i.e., class union) so Like many theoretical accounlts, Sidman's
that the context also drops out of the equiv- has evolved to accommodate empirical find-
alence relations (1994, p. 530). He also ar- ings that do not fit easily with his initial for-
gues that these overarching stimulus classes mulations. The result, described in great de-
may be prevented from forming in the first tail in his 1994 book, is an extraordinarily
place by the employment of different rein- ambitious revision of existing behavioral the-
forcers and defined responses for each con- ory in which the notion of equivalence be-
ditional relation (1994, p. 413). comes a core explanatory construct from
Having subjects name the stimuli within which is derived, not only success on match-
each conditional relation may, in Sidman's to-sample tests, transfer of function, and lin-
view, be another way of either preventing or guistic achievements, but also a range of what
breaking down all-inclusive equivalence rela- have hitherto been regarded as basic learning
tions. Thus, in the example given above, if phenomena including conditioned reinforce-
subjects name each of the Al and Bi stimuli ment and Pavlovian conditioning. The critical
as "X" and the A2 and B2 stimuli as "Y," this tests for any grand theory such as this, how-
should result in the establishment of the ever, are (a) how it relates to empirical data
AlBl and A2B2 equivalence relations be- and (b) whether it is internally consistent. We
cause, according to Sidman, the different contend that Sidman's theory has problems
naming responses "X" and "Y," like any oth- on both counts, only some of which we will
er differentiated response, facilitate class in- deal with here.
tersection (1994, p. 414). If equivalence is a given, why is it rarely, if
 19383711, 1996, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1901/jeab.1996.65-185 by Capes, Wiley Online Library on [26/05/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
NAMING AND OTHER SYMBOLIC BEHAVIOR 229
ever, found in nonverbal animals? On the one ertoire" would enable subjects to behave in
hand, Sidman professes not to know for that way. Nevertheless, whatever his precise
which animal species equivalence is a given explanation of verbal rules and their relation
but, on the other, his view that conditioned to equivalence, one can conclude from Sid-
reinforcement and Pavlovian conditioning man's account that there are at least two
are derivatives of equivalence implies that kinds of equivalence, namely, that which aris-
species that show evidence of such condition- es directdy from experience with contingen-
ing processes should also show equivalence. cies (i.e., contingency generated) and that
This, of course, would apply to a vast range which is established by verbal rules or names
of animal species, but for none of them, apart (i.e., rule governed or verbally controlled;
from humans, do we have reliable evidence Sidman, 1992, pp. 21-22, 1994, pp. 305-306;
of equivalence. see also Pilgrim & Galizio, 1995). We will re-
Nevertheless, for Sidman, equivalence is a turn to this distinction later.
given for the human species at least. But the Not only does it appear that equivalence is
empirical data show that success on equiva- not a given for all nonhuman species or even
lence tests is strongly related to the develop-
ment of linguistic skills; children lacking the for all individual humans, but for any partic-
necessary level of verbal skill fail these tests. ular human it does not appear to be a given
We have difficulty seeing how developmental at all times during testing. As several studies
differences like these can be explained by a have shown (see Sidman, 1994, p. 273), it is
theory that assumes equivalence as a given often the case that subjects initially fail to pass
for Homo sapiens in general. tests of equivalence but, as testing continues,
In view of the developmental data, it is par- performance indicative of equivalence rela-
adoxical that, as an explanation of why some tions emerges. Sidman suggests that this de-
adult human subjects fail tests of equivalence, layed emergence is due to interference from
Sidman has suggested that because equiva- other equivalence relations established out-
lence is a primitive stimulus function, it side of, or even within, the experimental set-
emerges first in early childhood and is only ting:
later broken down or circumscribed by verbal
rules (1992, p. 22). Once again there are no A well-designed test will arrange test trials so
experimental data to support this surmise; on that the experimentally established equiva-
the contrary, the available evidence indicates lence class provides the only basis for classifi-
that equivalence is absent early in human de- cation that remains possible-that "works"on
velopment but is facilitated by later language every trial. Thus, the grounds for relating the
learning. A further paradox, of course, is that sample and a comparison may vary from one
the domain most free of verbal rules, namely test trial to the next until finally, the subject
nonhuman animal behavior, is just where hits on a consistent basis-the experimentally
equivalence is most difficult to obtain. established equivalence relation that remains
Sidman's general treatment of verbal rules possible from trial to trial. (1994, p. 512)
appears to contain further inconsistencies. At
times he maintains that, when they are mem- But if, as is often the convention, testing takes
bers of an equivalence class, verbal rules are place in the absence of reinforcement or em-
no different in function from other responses
and stimuli and thus have no special media- pirical feedback of any kind, how is it that
tional role (1994, pp. 510-511), whereas at subjects can have a "classification" that can
other times he seems to ascribe a special role "work" or can vary their responses until they
to verbal repertoires. For example, in criticiz- "hit on a consistent basis"? (see Pilgrim,
ing Hayes' relational frame theory he writes, Chambers, & Galizio, 1995). This is just the
"I find it difficult to see how a common re- kind of behavior we would expect in organ-
sponse (equivalencing?) to exemplars that isms that are able to name the stimulus pairs
have nothing in common except the relation or formulate rules for responding to them
could arise in the absence of a highly com- that they can then proceed to combine and
plex verbal repertoire" (p. 556). But he does recombine until they formulate a consistent
not explain how such a "complex verbal rep- rule for responding on all of the test trials
 19383711, 1996, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1901/jeab.1996.65-185 by Capes, Wiley Online Library on [26/05/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
PAULINEJ HORNE and C. FERGUS LOWE
(or, as Saunders & Saunders, 1995,6 have sug- infants. The very experiments that gave rise
gested, until the experimenter ceases to pre- to the concept of equivalence now constitute
sent the test trials). Certainly, it is not clear problem cases or exceptions to the general
from Sidman's account how a nonverbal or- rule that it is overarching classes, composed
ganism or an organism behaving nonverbally of all the experimental stimuli that should
could engage in behavior of this kind. emerge, rather than the experimenter-de-
One of the most radical features of Sid- fined classes. Sidman's explanation offered
man's new account of the relations between for this paradox, including his hypotheses
events in an equivalence class is the removal about common reinforcers and responses
of the distinction between responses, includ- "dropping out" and competition between
ing naming and verbal rules, on the one the three-term unit and the overarching class,
hand, and stimuli, both discriminative and re- seems to us to be vague with respect to the
inforcing, on the other. This innovation was behavioral principles involved and lacking
prompted, in part, by experimental data empirical support. For a behavioral primitive,
showing that equivalence class membership from which so many other behavioral phe-
can include reinforcing stimuli (see Mc- nomena are supposedly derived, true equiv-
Ilvane, Dube, Klederas, de Rose, & Stoddard, alence is remarkably elusive.
1992). In our view, however, it gives rise to
serious and far-reaching conceptual and em- Hayes' Theory of Equivalence
pirical problems. At the most basic level, re- Hayes' relational frame theory has been
moval of the stimulus-response dichotomy ap- oudlined in several papers (e.g., Hayes, 1991,
pears to remove a distinction between 1994; Hayes & Hayes, 1989, 1992), so only a
behavior and environment upon which be- brief summary will be presented here. A ma-
havior analysis, as well as many other sciences, jor distinguishing feature of Hayes' view, in
are founded. It might be countered that, be- common with the present account but in con-
cause most of behavior analysis is based upon trast to that of Sidman, is that it explains suc-
the study of the behavior of nonhuman spe- cess on equivalence tests in terms of the prior
cies under contingencies of reinforcement learning history of the subject. For Hayes,
that do not give rise to equivalence relations this involves a history of what is termed arbi-
in those species, this issue does not arise ex- trarily applicable relational responding, which en-
cept in the case of species for which equiva- tails "responding to one event in terms of the
lence is a given (i.e., humans). But to argue other based on contextual cues to do so. It is
thus would be to severely limit the generality a pattern of the mutual transformation of
of Sidman's theory, and difficulties vis-a-vis stimulus functions" (Lipkens et al., 1993, p.
human behavior, particularly human lan- 204). The theory is based upon a few core
guage, would remain. These difficulties will ideas. Given that human and nonhuman spe-
be considered in detail below, following our cies can learn to respond to nonarbitrary re-
appraisal of Hayes' account of equivalence. lations between stimuli, it is argued that per-
In summary, what began as a description of haps some species can learn to respond to
novel behavioral relationships on match-to- relations between stimuli when these rela-
sample tasks has now evolved into a very gen- tions are not defined by the physical form of
eral theory that embraces most aspects of be- the stimuli but by contextual cues. Because
havior analysis. It appears to us, however, that only contextual cues are required, such rela-
as the scope of the theory has extended, the tional responding is held to be arbitrarily ap-
theory has become increasingly removed plicable to any event. An analogy is drawn
from empirical foundations. Equivalence has with the findings of the literature on gener-
become a behavioral primitive that is not alized imitation (e.g., Gewirtz & Stingle,
found where we should most readily expect 1968) that indicates that organisms can learn
it, namely, in nonhuman animals and human overarching behavioral classes. The action of
relating arbitrary stimuli in equivalence tasks,
6Saunders, R. R., & Saunders,J. M. (1995, April). The it is proposed, may be a similarly learned
roles of generalised conditional responding and chance in the overarching class of instrumental behavior,
emergence of equivalence-indicative performances. Paper pre-
sented at the annual meeting of the Experimental Anal- which Hayes (1992, p. 110) terms equivalenc-
ysis of Behaviour Group, London. ing.
 19383711, 1996, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1901/jeab.1996.65-185 by Capes, Wiley Online Library on [26/05/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
NAMING AAD OTHER SYMBOLIC BEHAVIOR 231
Arbitrarily applicable relational responding could derive the frame of coordination and
has three primary properties. The first of show equivalence classes." And yet again,
these is mutual entailment and involves re- Hayes (1991, p. 25) claims that equivalence
sponding to one event in terms of the other, emerges because "mutual entailment, com-
and vice versa. So if A is better than B, it is binatorial entailment, and transfer of func-
mutually entailed that B is worse than A. Mu- tion are directly trained." This latter view is
tual entailment is the generic case of what is elaborated by Hayes and Wilson (in press),
termed symmetry in the equivalence literature who suggest that this might involve, with dif-
(Hayes, 1994, p. 11) in which A is the same ferent combinations of stimulus pairs, the di-
as B and, through mutual entailment, B is the rect reinforcement of symmetry, transitivity,
same as A. The particular relation (e.g., same- and equivalence, they also suggest that rein-
ness, opposition, distinction, etc.) between forcing symmetry might be enough to give
events is controlled by contextual stimuli. rise to equivalence but that "some small
The second is combinatorial mutual entailment, amount of training in combinatorial entail-
according to which if A is related in a partic- ment will probably also be needed." Al-
ular way to B and B is also related to C, then though Hayes (1994) favors the last of these
(in addition to the mutual entailment of re- three possibilities, the theory itself does not
lations between A and B and B and C) a com- indicate why one rather than another of these
binatorial derived relation is mutually en- histories would be sufficient to yield a frame
tailed between A and C. Combinatorial of coordination (i.e., equivalence).
entailment is the generic case of what are How does the history work? This difficulty aris-
termed transitivity and equivalence in the es from the fact that the theory does not
equivalence literature (see Hayes, 1994, p. clearly specify what are the behavioral prin-
11). The third property is transformation of ciples that govern the establishment of rela-
stimulus function, which specifies that in a giv- tional frames. An example of the kind of his-
en context, if A is related (e.g., by opposi- tory that should bring about symmetry or
tion) to B and, in addition, A is given a non- mutual entailment is held to be one in which
relational stimulus function, then this will an organism is presented with sample-com-
result in a derived nonrelational stimulus parison stimuli such that the A -> B relation
function for B, in accord with the specified is reinforced, and then the reversal B -* A,
relation. The nonrelational functions of A then C -> D and D -4 C, and so on. At some
that are derived for B will also be under con- point, it is argued, the organism, when
textual control. trained on X -* Y, will show the derived re-
The term relational frame (or framing rela- lation Y -* X; thus generalized symmetrical
tionally) refers to arbitrarily applicable rela- responding has been established. Apart from
tional responding that shows the three qual- characterizing this behavior as an overarch-
ities oudlined above. According to Hayes ing response class that occurs in the presence
(1994, p. 28), one such frame, that is, the of contextual cues (as is, of course, true of all
frame of coordination or sameness, establishes operant behavior), and appealing to an anal-
equivalence classes. ogy with generalized imitation, Hayes pro-
What is the history ?Although the theory pro- vides no further description or functional
poses that subjects require a history of ex- analysis of the behavior involved. The prob-
amples of reinforced relational responding in lem with the imitation analogy is that imita-
order to abstract a relational frame and show tion does not have any of the defining fea-
equivalence, precisely what this history tures of relational framing (see Skinner,
should be is not clear. Hayes (1989, p. 391) 1953, pp. 119-122) and hence does not clar-
argues that one possible approach with non- ify the behavioral principles involved in ab-
humans may be "to provide an extensive re- stracting a frame of coordination or sameness
inforced history with symmetrical relations" from a history of reinforced reversals of stim-
after which equivalence might emerge. On ulus pairs. This is not, of course, to assert that
the other hand, Hayes and Hayes (1989, p. success on tests of equivalence cannot ever be
174) propose that "a child given only a his- brought about by histories of relational re-
tory of arbitrary matching-to-sample that re- sponding which do not involve naming, but
inforced symmetry, reflexivity and transitivity, merely that, were such a phenomenon ever
 19383711, 1996, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1901/jeab.1996.65-185 by Capes, Wiley Online Library on [26/05/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
232 PAULINEJ. HORNE and C. FERGUS LOWE

to be convincingly demonstrated in a manner lations solely on the basis of reinforced rever-

that ruled out known behavioral principles, sals of stimulus relations. Second, in this par-
relational frame theory could provide little by ticular case, consider what happens when,
way of explanation (see Sidman, 1994, p. after a history of naming acquisition, a new
557). Name Z -* Object Z relation is trained: The
Naming as the paradigm case? In all of Hayes' child hears /Z/ -* orients to and sees Z
writings, the only specific example of a his- (trained); the child sees Z -> says "Z"
tory that is given from real life to account for (emerges). However, according to relational
the initial establishment of any relational frame theory and the notion of generalized
frame is for the frame of coordination. This symmetrical responding, what should emerge
example is described in a number of papers when the child sees Z is her hearing /Z/, not
(Hayes, 1991; Hayes & Hayes, 1989, 1992; saying "Z." How the child comes to say "Z"
Lipkens et al., 1993; Steele & Hayes, 1991) when she sees the object and what happens
and interestingly, from our perspective, it to the predicted hearing /Z/ response are un-
concerns the acquisition of naming in young avoidable questions here. To deal with these
children. The account runs as follows: When problems, Hayes and colleagues (Lipkens et
the young child is presented with an object al., 1993, p. 216; Steele & Hayes, 1991, p.
(Object A) and asked "what's that?" correct 553) have proposed that when presented with
responses (i.e., the child saying "A") are re- the object and asked "what's that?" the child
inforced (with smiles, etc.). This establishes first hears /Z/, thereby conforming to what
the Object A -e Name A relation. However, would be expected from relational frame the-
the child is also often asked "where's A?" and ory, and then utters the name "Z." It follows
presented with several objects, including A. from this account that when a child sees an
When the child selectively orients or points object and emits its name (i.e., that part of
to A, a social reinforcer is again given, there- the name relation that Skinner described as
by training the relation Name A -* Object A. the tact), the utterance is not controlled by
Following repeated learning of object -o the object itself but by a hypothesized hearing
name and name -> object relations for dif- response that precedes it. No rationale is pre-
ferent object-name pairs, when a new Object sented to support this notion of children
X - Name X relation is trained, the Name hearing names before they can utter them; it
X - Object X relation emerges or is derived. certainly does not feature in Skinner's ac-
Similarly, when a new Name Y - Object Y count of verbal behavior, nor, as far as we are
relation (i.e., hears /where's Y?/ - orients to aware, is there any evidence in the develop-
Y) is trained, the Object Y -> Name Y relation mental literature that this occurs. It is true
(i.e., sees Y - says "Y' in response to "what's that children hear their own utterances, a be-
that?") emerges. Such symmetrical respond- havior that features prominently in our ac-
ing, it is held, occurs in particular contexts, count of naming acquisition, but this occurs,
the context here being naming indicated by of course, after they have spoken and not be-
cues such as "what's that?" "In short, with fore, as the relational frame account requires.
enough instances of directly trained symmet- Third, although the above example shows
rical responding, symmetrical responding that naming acquisition cannot be described
may emerge with respect to novel stimuli in as mutual entailment, neither does it involve
that context. That is, the extensive training combinatorial mutual entailment; on both
history may be brought to bear by a given these grounds naming fails to meet the cri-
context and provide a basis for generalized teria for participation in a relational frame.
symmetrical responding" (Hayes & Hayes, There thus appears to be a major inconsisten-
1989, p. 168). cy at the heart of the account: Either the his-
Although we would clearly not doubt the tory involved in naming acquisition is suffi-
importance of naming in bringing about suc- cient for the child to derive a relational
cess on tests of symmetry and equivalence, frame, in which case combinatorial entail-
this example is problematic for the relational ment (i.e., transitivity) is not necessary for
frame account. First, there is the general framing (see Boelens, 1994), or naming, as
problem to which we have already alluded, described, is not in the frame of coordination
namely, how one can account for derived re- nor indeed in any other frame.
 19383711, 1996, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1901/jeab.1996.65-185 by Capes, Wiley Online Library on [26/05/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
NAMING AND OTHER SYMBOLIC BEHAVIOR 233
Far from providing a clear demonstration bal rules can help to form relational networks
of how generalized symmetry is established, and equivalence, although this is in turn ex-
the paradigm case of naming serves only to plained "on the basis of names as contextual
raise further questions about the theoretical cues for relational responses, and on the basis
coherence and applicability of relational of derived relations formed to the names
frame theory. It leaves unanswered questions themselves" (Hayes, 1994, pp. 19-20). Thus,
about which specific learning histories give as was the case with Sidman's account, we
rise not only to mutual entailment but also may conclude that relational frame theory
the frame of coordination and equivalence it- also allows for at least two basic kinds of
self. Indeed, nowhere is an example given of equivalence (i.e., one directly contingency
a history that might establish in young chil- generated, the second rule governed or oth-
dren or nonverbal animals any other relation- erwise verbally controlled). If contingency-
al frame, such as opposition or comparison, generated equivalence and related phenom-
nor are there empirical data from these sub- ena were ever reliably found in any nonverbal
jects that might support such hypotheses. species, perhaps after a history of reinforced
It is, of course, always a possibility that oth- " reflexivity," "symmetry, " " transitivity, "
er histories, perhaps either provided to non- "equivalence," or some combination of any
verbal organisms or not involving naming, or all of these, then, as Boelens (1994) has
might give rise to generalized symmetrical re- suggested, the concepts of generalized sym-
sponding and hence to success on match-to- metry, generalized transitivity, or generalized
sample tests of symmetry and equivalence. equivalence might be invoked by way of ex-
This has indeed been claimed by Schuster- planation. This notion of generalization
man and Kastak (1993) in a study of a sea would at least be more parsimonious than the
lion (but see above). On the other hand, ac- complex theoretical constructions of relation-
cording to relational frame theory, one might al frame theory, although in order to go be-
expect success in nonhuman animals (a) that yond mere redescription of the behavioral
are very closely related to the human species data, it would itself require explanation in
and (b) that have very extensive histories of terms of known or novel behavioral princi-
reinforced reversals of relations between stim- ples.
uli. Dugdale and Lowe (1990) tested this hy- However, until such phenomena are relia-
pothesis with 2 "language-trained" chimpan- bly found (if they ever are) with other spe-
zees, Sherman and Austin, that had had more cies, we are left with the findings from human
extensive learning histories, including histo- subjects who pass tests of equivalence. We
ries of reinforced reversals, than almost any have shown in the early part of this paper
other nonhuman animal (for details of their how, in learning to name, humans learn to
learning histories, see Rumbaugh, 1977; Sav- respond to arbitrary classes of stimuli. Simple
age-Rumbaugh, 1986). An extensive program name relations, more complex name se-
of controlled experimentation was carried quences, or verbal rules are all sufficient to
out using a visual-visual match-to-sample task enable subjects to pass match-to-sample tests
similar to that employed by Lowe and Beasty of equivalence. Is any other concept or prin-
(1987). In spite of hundreds of training and ciple needed to account for such behavior in
humans? Has there ever been an instance of
test trials, including blocks of reinforced test equivalence in humans that has been directly
trials, neither of the 2 chimpanzees showed contingency generated? Until there is evi-
any evidence of success on tests for symmetry. dence of the latter, it would certainly be more
In any case, if a pigeon, a chimp, or a sea parsimonious to adopt the naming account
lion were to pass such tests for symmetry or rather than one that must invoke the several
equivalence, what would it tell us about (a) conceptual layers of relational frame theory
how they passed the test (i.e., the behavioral to explain performance on match-to-sample
principles involved), (b) whether such prin- tests.
ciples also govern the performance of hu-
mans who pass equivalence tests, and (c) Equivalence, Relational Frames, and
what, if anything, this has to do with language Language
and naming. Within the Hayes account there According to the Hayes account, relational
is an acknowledgment that naming and ver- frames are the defining characteristic of ver-
 19383711, 1996, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1901/jeab.1996.65-185 by Capes, Wiley Online Library on [26/05/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
234 PAULINEJ. HORNE and C. FERGUS LOW%E
bal events (Hayes, 1994, p. 12). Wulfert and relation-sees Object Z, says "Z"-has
Hayes (1988) maintain that the symmetrical emerged (see pp. 199-209, Naming).
relations among stimuli established by fram- Indeed, although we have concentrated so
ing provide the basis for what they term ref- far on the acquisition of naming in child-
erential meaning. They write, "The word is a hood, at any level of analysis the assumption
symbol for the referent and the referent is that the relation between names and objects
the meaning of the word because both are is symmetrical (Dugdale & Lowe, 1990)
members of the same equivalence class. In seems, upon scrutiny, to be without founda-
this sense, stimulus equivalence transforms tion. We might say that "X" refers to or is the
nonlinguistic conditional discriminations into name of Object X. But we cannot, as true sym-
semantic process" (p. 126). Similarly, Sidman metry would demand, say the reverse, that is
has argued that stimulus equivalence is "a lin- that Object X refers to or is the name of "X."
guistic prerequisite" (1986, p. 226) and that Thus, for example, although the name
"words are equivalent to their 'referents'" "chair" refers to a class of objects with par-
(Sidman et al., 1986, p. 2). He observes, "By ticular features, these objects either singly or
reacting to a word as to an equivalent stimu- in groups do not refer to the name "chair."
lus-the meaning of the word-a person can In brief, the relation between a name and
behave adaptively in an environment without that which it names is fundamentally asym-
having previously been exposed to it" (1986, metrical.
p. 236). As both Hayes and Sidman have rec- Sameness, meaning, and naming. The view
ognized here, the basis for any conception of that equivalence incorporates the concept of
reference or meaning is the relation of words sameness may have helped to give rise to the
or names to objects and events, and clearly notion that equivalence is a linguistic prereq-
both accounts assume that this relation is cap- uisite. This is certainly clear in relational
tured by the notion of equivalence. This is a frame theory, in which the most fundamental
questionable assumption. frame is held to be that of coordination or
Symmetry and naming. First we shall consider sameness, which is said to be abstracted by
whether the relation between a name and the the child from early language training of the
kind described above and which enables new
object named (i.e., the referent) is one of name relations to emerge. Similarly, Sidman
symmetry. The example given by Hayes of has proposed that "sameness is a prerequisite
naming acquisition in the young child, which for equivalence. Therefore ... it is also a pre-
we have considered in some detail above, pro- requisite for the emergence of simple mean-
vides excellent evidence to the contrary. This ings, vocabularies, or 'semantic correspon-
interaction, although indeed typical of what dences'" (1986, p. 227). Further pursuing
happens as relations are established between this notion, Sidman argues that subjects who
children's utterances and their other behav- show equivalence between particular words
ior and objects in the environment, cannot (e.g., "Route 128" on a map) and objects
be described as an example of symmetry. It (e.g., the road on which they are driving)
is, however, consistent with the account of might be expected to treat them as the same
naming provided in the first part of this pa- (e.g., by driving their car onto the words).
per. The problem is to explain the fact that, This does not happen, he argues, only be-
having been taught to look at Object Z when cause we learn to circumscribe or inhibit the
she hears /Z/, the child says "Z" when she primitive stimulus function of equivalence
next sees the object. We propose that any ac- that would otherwise impel us to show such
count of how this novel utterance occurs behavior. We really would, for example, eat
must recognize, as the Hayes view of symme- the word "bread," did we not learn through
try does not, the role of echoic behavior and experience or through rules that words, even
the importance of the child learning to echo when "equivalent" to foods, are not edible.
the utterances of others when asked, for ex- Thus, it is assumed in both the Sidman and
ample, "where's the 'Z'?" If the child echoes Hayes accounts (a) that sameness can de-
"Z" when she hears /Z/ in the presence of scribe the relation between certain names
Object Z, then seeing the object will come and objects and also (b) that it is a prereq-
directdy to control her saying "Z." The new uisite for successful performance on match-
 19383711, 1996, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1901/jeab.1996.65-185 by Capes, Wiley Online Library on [26/05/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
NAMING AND OTHER SYMBOLIC BEHAVIOR

to-sample tests. But a name is never the same 1972, p. 170), that gives rise to what equiva-
as the object named. To say of a chair that it lence researchers and others have termed se-
is "furniture" does not entail that "furni- mantic relations, referential meaning, and vocab-
ture" is a chair or that furniture is the same as ularies (see pp. 213-215, Naming and Symbolic
a chair. Similarly, to also call tables "furni- Behavior).
ture" does not entail that chairs and tables Sameness and equivalence. As for sameness
are the same. Although different objects with being a prerequisite for equivalence, if a sub-
a common name (e.g., "chair") may be treat- ject, using the common name "furniture" on
ed in the same manner or given a similar a match-to-sample task, is shown to have es-
function (e.g., by being sat upon), as Skinner tablished an equivalence class of chairs, ta-
(1957, p. 87) observed, we do not behave to- bles, and chests of drawers (or their pictures),
ward the name as we do toward the named this clearly does not mean that the relation
object or event. We do not sit on the printed formed was one of sameness. Nor, of course,
word CHAIR when we see it, nor can we sit does it mean that because they were in an
on the spoken word "chair." Pace Sidman, it equivalence relation, the chair was the name
stretches credibility to maintain that we have of a table or that either was the name of a
primitive tendencies to do so. To argue that chest of drawers. To take an example from
"the referent is the meaning of the word" the experimental literature, when stimuli
(Wulfert & Hayes, 1988, p. 126) or that "a within experimenter-defined classes in match-
computer is a 'computer'" (Hayes & Hayes, to-sample experiments are abstract shapes
1989, p. 169) is thus to fundamentally mis- (e.g., a triangle, a green square, and a vertical
construe word meaning and reference; any line), there is no reason to assume that any
particular referent, such as a particular com- of the visual stimuli themselves are treated by
puter, is just one of an indefinite range of the subject as being the same as, or indeed
objects to which the name "computer" ap- the name of, any other stimulus; intraverbal
plies. To even claim that that class of objects naming (e.g., "up-green-triangle") or verbal
"is the meaning" of the word "computer" is rules (e.g., "goes with") that have nothing to
laden with further logical and conceptual do with the concept of sameness may be op-
problems (Hunter, 1974; Ryle, 1949; Skinner, erating.
1957, pp. 86-89). Further experimental evidence that sub-
We believe that the account of naming ac- jects do not invariably treat stimuli within an
quisition presented in the early part of this equivalence class as the same or equally sub-
essay avoids many of these pitfalls. It does not stitutable comes from studies conducted by
view names as being the same as the objects Fields and colleagues (Adams, Fields, & Ver-
or events named or as being equal to them, have, 1993; Fields, Adams, & Verhave, 1993).
nor does it consider that objects are part of These have shown that such stimuli can oc-
names or even part of the meaning of names. casion very different kinds of behavior (e.g.,
Naming is behavior and, like any other be- as indicated by reaction-time measures, errors
havior, it occurs in relation to objects and in test trials, transfer of novel responding
events but it is not to be confused with them. within the class, and within-class stimulus
To blur this distinction or, indeed, to remove preferences) depending upon their distance
it entirely (see Sidman, 1994, p. 386) is to from the nodal stimulus and the direction of
introduce myriad problems. We have already training of the stimulus relations. Thus, al-
shown how the young child learns a name though according to the logico-mathematical
(e.g., "furniture") that relates to a class of criteria of stimulus equivalence, stimuli may
objects (e.g., chairs, tables, etc.) that are phys- appear to be related to each other on the
ically disparate. The child's use of the con- basis of equality (Sidman, 1990), more de-
ventional name "furniture" for each of these tailed analysis of behavior reveals substantial
objects establishes the conventional stimulus differences in subjects' responses to them
class and governs her use of each class mem- (Fields et al., 1993).
ber (e.g., the placing of it in a dolls' house). In defense of Sidman's account, it should
It is this behavior, or as Wittgenstein would be noted that, in his critique of relational
have it, the use of naming in "the actions into frame theory (Sidman, 1994, p. 559), he has
which it is woven" (Wittgenstein, 1953, p. 5, argued that equality is only one example of
 19383711, 1996, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1901/jeab.1996.65-185 by Capes, Wiley Online Library on [26/05/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
PAULINEJ HORNE and C. FERGUS LOWE
an equivalence relation. Equivalence, he the contrary, a great deal of experimental ev-
maintains here, is more general and includes idence showing major differences between
relations such as "is parallel to" and "is con- contingency-shaped and rule-governed be-
gruent to" (for triangles in a plane) and "has havior (Lowe, 1979, 1983) that is entirely
the same teacher as." This view, however, rais- consistent with Skinner's observation that
es further questions, namely, (a) how can it
be reconciled with the position that sameness If cognitive psychologists were correct in say-
is a "prerequisite for equivalence" (Sidman, ing that rules are in the contingencies, it
1986, p. 227), and (b) given that it is not the would not matter whether we learned them
from the contingencies or from the rule-in
formal characteristics of the stimuli that de- other words, from acquaintance or descrip-
termine performance on the equivalence tion. The results, however, are obviously dif-
test, how does a nonverbal organism or an ferent. ... There is a difference because rules
organism behaving nonverbally acquire such never fully describe the contingencies they are
relations as "has the same teacher as"? designed to replace. There is also a difference
Equivalence and rule-governed behavior Ac- in the states of the body felt. (1989, pp. 43-
cording to both Sidman and Hayes, rule-gov- 44)
erned behavior is based upon equivalence re-
lations, and this, they believe, deals with one So once again, there is the danger that the
of the shortcomings of Skinner's account, notion of equivalence may mask rather than
that is, that it does not provide a behavioral clarify crucial behavioral distinctions. In us-
specification of how rules "specify" contin- ing the terms describe and specify for the rela-
gencies. As Sidman has observed, "If equiva- tion of rules to contingencies, Skinner may
lence gives rise to rules, then for a rule to have been closer to the mark than his critics
specify a contingency may simply mean that were. Nevertheless, it must be conceded that
the rule and the contingency are members of these terms do require more precise behav-
an equivalence class" (1990, p. 106; see De- ioral specification, and it is this that we have
vany et al., 1986; Hayes & Hayes, 1992). What aimed to achieve here. Certainly, little prog-
is not clear from either the Sidman or Hayes ress will be made in understanding how rules
accounts, however, is how rules and contin- come to govern behavior until we understand
gencies can be members of the same equiva- the role of the verbal units of which rules are
lence class or how such a relation could come composed, how they come into being and
about. For, in the same way that we have dem- how, either individually or in combination,
onstrated that names and objects are not sym- they affect other behavior. This has so far not
metrically related, it can be shown that the been attempted in the equivalence literature
relation between rules and contingencies is and although we have attempted to lay the
not one of either symmetry or equivalence. foundations for such an understanding, there
Under certain circumstances verbal rules, is clearly much more to be done before a
perhaps acquired by instructions from others, comprehensive analysis can be provided.
govern behavior but this behavior does not Summary. From the evidence presented in
govern the verbal rules. Indeed, although this section there are a number of conclu-
there may be an interaction, it is a well-estab- sions that can be drawn on the relation be-
lished finding that rules can be remarkably tween equivalence and language. (a) The re-
insensitive to the consequences of the behav- lation between name and objects is not one
ior they govern (Catania et al., 1982; Lowe, of symmetry or equivalence. The fact that
1979, 1983). It makes sense to talk, as Skinner subjects pass match-to-sample tests cannot be
does, of rules that describe or specify contin- taken to indicate that sameness is involved in
gencies but not of contingencies that specify their behavior; nor can we assume that the
or describe rules. Thus, it is not the case that members of an equivalence class are the
rules are the "same as" or "equal to" contin- names of other members, or refer to, sym-
gencies, or, as Sidman (1994) would have it, bolize, or mean them. Although match-to-
that the distinction between rules and contin- sample performance and the construct of
gencies, like that between other responses equivalence cannot provide a model for lin-
and stimuli, breaks down if both participate guistic behavior it may, however, be diagnos-
in an equivalence class (p. 386). There is, on tic of its effects, that is, success on symmetry
 19383711, 1996, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1901/jeab.1996.65-185 by Capes, Wiley Online Library on [26/05/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
NAMING AM) OTHER SYMBOLIC BEHAVIOR 237
and equivalence tests may be a good indicator ied and fundamental verbal processes. Most
that the subject has acquired naming. researchers in this area, although ostensibly
investigating the "new" phenomenon of stim-
The Construct of Stimulus Equivalence: ulus equivalence, may in fact have been study-
Do We Need It? ing naming and other forms of verbal behav-
Problems with the unitary concept of equiva- ior, but under a different name. Their
lence. We have shown that success on match- tendency to ignore the possible role of sub-
to-sample tests, and thus, according to its op- jects' verbal behavior and hence not even to
erational definition, stimulus equivalence, record it (but see Dugdale & Lowe, 1990;
can be established by several different behav- Saunders & Spradlin, 1990; Wulfert et al.,
ioral routes, including common naming, in- 1991) may have led many researchers to over-
traverbal naming, and more complex verbal look significant variables that influence test
rules such as "A goes with B," "A is the same outcome. This seems to be a good example
as B," and so on. It has also been suggested, of what Skinner (1969) has termed the for-
as a theoretical possibility, that there might malisticfallacy (see also Vygotsky, 1978, p. 62),
be a type of equivalence that is directly con- whereby an undue emphasis is placed upon
tingency generated. But these various paths the formal characteristics of behavior (e.g.,
to success represent very different behavioral success on equivalence tests) at the expense
processes that, although they may yield the of an analysis of controlling relations (e.g.,
common outcome of correct performance on the role of verbal stimuli).
a particular match-to-sample configuration, Whether or not subjects succeed on equiv-
may have other differing behavioral out- alence tests is not a matter of a straightfor-
comes including, we predict, different per- ward logical or mathematical relation but
formances on tests of stimulus class exten- rather of a behavioral process that varies with
sion. Evidence to this effect comes from a a number of factors. These include whether
study by Bentall et al. (1993), who have the subject produces names in ways that will
shown that adults taught a common name for generate stimulus class categories in line with
stimuli later related in a match-to-sample task those specified by the experimenter. And this
showed no reaction-time differences to the in turn, as we have seen, is determined by a
stimuli on baseline, symmetry, transitivity, and variety of other factors including whether the
equivalence trials. When, however, subjects procedure is audio-visual or visual-visual,
were taught individual stimulus names rather whether names are provided directly by the
than common names, they showed longer re- experimenter or generated by prior history
action times on the transitivity and equiva- of stimulus classification, and, of course,
lence trials than on the symmetry and base- whether the names thus generated control
line trials. These results indicate that the appropriate listener behavior of pressing
although, where meeting the mathematical the corresponding response keys. If we are
criteria for equivalence is concerned, the be- interested in establishing the determinants of
havior of subjects with common naming may human match-to-sample performance, future
differ little from that of those with intraverbal research should explore these relations in de-
or other strategies, when other measures such tail.
as reaction times are recorded, there is evi- Given, as other investigators have also sug-
dence of behavioral differences (Adams et al., gested, that the existing mathematical ap-
1993; Fields et al., 1993). proach to defining equivalence is problem-
Different verbal repertoires can have the atic (Pilgrim & Galizio, 1995; Saunders et al.,
common function of generating success on 1993; Saunders & Green, 1992), as are at-
equivalence tests, and it is easy to see how this tempts to characterize it in terms of simple
fact could have given rise to usage of a com- notions such as sameness, one way forward
mon term, stimulus equivalence, to describe might be to create a new definition. We sug-
this success and to a belief in a correspond- gest, however, that the flight to logico-math-
ingly new phenomenon. What has not until ematical models is a diversion from the cen-
now been appreciated, however, is that suc- tral task of conducting a functional analysis
cess on stimulus equivalence tests may be a of the conditions that give rise to success on
secondary and indirect outcome of more var- matching tests so that these definitional issues
 19383711, 1996, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1901/jeab.1996.65-185 by Capes, Wiley Online Library on [26/05/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
238 PAULINEJ. HORNE and C. FERGUS LOWE
can be resolved, including the question of structs from relational frame theory required:
whether a definition of stimulus equivalence arbitrarily applicable relational responding,
is required at all. If success on tests of stim- relational frames, the frame of coordination,
ulus equivalence is indeed a by-product or mutual entailment, combinatorial mutual en-
secondary outcome of naming and other ver- tailment, transformation of stimulus func-
bal processes, the very notion of a "general- tion, and equivalence? Or are any or all of
ized concept of equivalence" (Johnson & Sid- the Sidman hypotheses required? His theory
man, 1993, p. 346) begins to evaporate. As also views naming as being derived from the
we have shown, we do not need it to account construct of equivalence but in an even more
for the success of human subjects on match- complex manner, with much of the theoreti-
to-sample tests or to account for the range of cal effort (see Sidman, 1994, pp. 389, 554)
other emergent behavior that occurs during invested in explanations of why equivalence
early language development in the child. In- does not occur rather than in a functional
deed, the arbitrary match-to-sample proce- analysis of how both it and naming come
dure itself (as distinct from identity match- about.
ing), particularly visual-visual matching, is a As Skinner (1950) pointed out, another
highly artificial set of circumstances of a sort problem with some kinds of theory is that
infrequently, if ever, encountered by most when the theory is overthrown, much of the
children and thus is an odd experimental associated research is discarded. This has
paradigm upon which to base a general un- happened once before with a research enter-
derstanding of new or derived relations or prise that was based on the notion of equiv-
language itself. alence and that also claimed that equivalence
Are theories of equivalence necessary? As we was at the center of linguistic behavior (Jen-
have attempted to show, arbitrary stimulus kins, 1965). The search for the key to equiv-
classes, upon which the concept of stimulus alence failed then, the theory faded away, and
equivalence is based, are a fundamental as- the research seems, indeed, to have been
pect of the name relation, hence the rele- largely forgotten. This time around, should
vance of equivalence research for the analysis not our research enterprise deal directly with
of verbal behavior. But is the notion of stim- verbal behavior itself unmediated by the con-
ulus equivalence a helpful explanatory device struct of equivalence? Verbal behavior will
or does it introduce conceptual confusion? not fade away.
Stimulus equivalence, as originally defined, To study naming directly entails, as we have
describes a set of behavioral relations on indicated, the experimental investigation,
match-to-sample tasks. To label these behav- from birth, of how the young child learns the
ioral outcomes as equivalence, symmetry, or re- individual behavioral relations that in com-
lational frames and then to use these bination bring about naming. This approach
constructs to account for the behavior from would certainly be more parsimonious; it is
which they were derived together with a also in the best tradition of behavior analysis.
range of other linguistic and transfer-of-func- Such a study would enable researchers to
tion phenomena requires justification. One come to terms with the full complexity of the
negative effect may be that, in erecting an phenomenon, both in terms of the condi-
elaborate set of abstractions upon this limited tions that give rise to it and the interactions
behavioral base, a false sense of security is cre- between multisensory stimulation and multi-
ated and it is concluded that language acqui- modal responding that it entails, including
sition, verbal rules and, of course, derived emotional behavior and the effects of classi-
stimulus classes themselves have been ex- cal conditioning. Such complexity cannot be
plained. encompassed by the logico-mathematical ab-
A more serious difficulty arises, however, stractions of equivalence.
when theoretical constructs introduce con- It is researchers' recognition of the impor-
ceptual confusion and impede research. tance of emergent behavior that has been re-
Again we have indicated how some constructs sponsible for much of the interest in stimulus
in the equivalence literature might have this equivalence. The present account suggests
effect. For example, in order to account for that their work on the latter should now lead
naming, are any or all of the following con- to examination of the variables that are re-
 19383711, 1996, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1901/jeab.1996.65-185 by Capes, Wiley Online Library on [26/05/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
NAMING AMD OTHER SYMBOLIC BEHAVIOR 239

sponsible for bringing about "new" behavior question then would become how to account
and the direct study of language acquisition for the verbal behavior.
and its effects on other behavior. In switching This focus would help to remove some of
the focus of research from what may appear the other incompatibilities. For example, one
to many to be an arcane phenomenon called aspect of Sidman's (1994) recent account
stimulus equivalence to phenomena such as that seems most likely to be contentious and
naming and its relation to categorization, we difficult to reconcile with our view is the no-
can make common ground with other re- tion that basic learning phenomena, such as
searchers both in psychology and other dis- Pavlovian conditioning and conditioned re-
ciplines who themselves recognize these lat- inforcement, are derived from stimulus equiv-
ter areas of inquiry as valid and important. alence. But if equivalence is verbally driven
Furthermore, we are likely to enhance the and occurs only in humans, then the notion
practical applications of research in the field. that naming enters into the establishment of
Instead of assuming that emergent behavior classical conditioning effects and conditioned
comes about as a result of an innate unana- reinforcement would not be at all problem-
lyzable stimulus relation best revealed in the atic; indeed, there is considerable experimen-
match-to-sample paradigm, practitioners will tal evidence to support such a view (e.g.,
advance towards understanding how lan- Sokolov, 1972). That said, we would, never-
guage itself emerges and how it can be used theless, have to maintain that these condi-
to foster the development of stimulus classes tioning phenomena are not invariably de-
and, hence, an almost infinite range of new rived from naming or equivalence but,
behavior. indeed, are among the true behavioral prim-
itives that give rise to verbal behavior and
Reconciling the Different Accounts stimulus classes in humans.
Although in the foregoing we have stressed Shifting the emphasis to verbal behavior
the differences between our account and should also help to establish a greater com-
those of Sidman and Hayes, all three ac- patibility between the Hayes account and our
own. Both accounts assume that the phenom-
counts share common ground. First, there is ena of interest come about as a result of a
the recognition that it is important to estab- learning history, which even in the Hayes ac-
lish how subjects, without having been direct- count is recognized as "lying largely in the
ly trained to do so, can in some contexts treat context of language training" (Steele &
structurally different stimuli as if they were Hayes, 1991, p. 20). Relational frame theory
interchangeable, that is, as members of a sees the repeated reinforcement of relations
stimulus class. Second, there is the conviction between stimulus pairs and their reversals as
that the phenomena studied are closely sufficient, whereas we maintain that, in ad-
bound up with linguistic behavior. In view of dition to the stimuli presented, what the child
these central similarities, might not the dif- says, what she hears, and the listener behavior
fering accounts be reconciled in order to ad- thus engendered are all critical elements in
vance a common research agenda for the fu- acquisition of the crucial repertoire. Thus,
ture? Perhaps one way of achieving this the accounts are similar in their general em-
reconciliation would be to recognize that phasis on learning history, although they car-
there might be at least two ways that a subject ry different implications for how naming
could achieve success on an equivalence test, should be taught and how it enters into re-
that is, by way of either directly contingency- lations with other behavior. Nevertheless, it
generated behavior or rule-governed or ver- should be possible to submit the different hy-
bally controlled behavior. The principles un- potheses to experimental test on this issue so
derlying the former might not be clear, but that a common basis could be established
it would be an empirical question as to wheth- from which to move forward. If, of course, it
er it does or does not occur. If it could not was found that very young infants, prior to
be found (e.g., in nonverbal species or non- experiencing either of the kinds of history
verbal infants) then perhaps all three ac- proposed in this or in the Hayes account,
counts would have to accept that we must could readily form equivalence classes, which
deal with a verbally driven phenomenon. The in turn facilitated acquisition of naming, then
 19383711, 1996, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1901/jeab.1996.65-185 by Capes, Wiley Online Library on [26/05/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
240 PAULINEJ. HORNE and C. FERGUS LOWE
both the present account and that of Hayes governance, and human consciousness. Referring
would need to be revised in line with Sid- to the special accomplishments of verbal be-
man's theory. We have reviewed the existing havior, Skinner (1957) says that the speaker
evidence bearing on these issues in earlier is "a locus-a place in which a number of
sections, and the reader can judge according- variables come together in a unique conflu-
ly, but much of the critical experimentation ence to yield an equally unique achievement"
remains to be done. (p. 313). We would add that it is only when
the speaker becomes a speaker-listener and
can name that this unique achievement is re-
CONCLUSION alized.
The starting point for the present paper An effective behavioral theory of language
has been Skinner's account of verbal behav- acquisition should be of benefit not just out-
ior, which, although central to his thinking, side the behavior-analytic tradition but also
has not had nearly the impact or recognition within, because, as some have argued (e.g.,
within psychology and other disciplines that Richelle, 1993), the topic, centrally important
many, including the present authors, main- though it is, has so far not received adequate
tain that it deserves (Andresen, 1990; Ri- attention from behavior analysts themselves.
chelle, 1993). Much of the important re- As we have stressed, the account presented
search on children's language acquisition here is a theoretical one, and the particular
conducted in recent years, for example, behavioral relations we have oudlined, begin-
makes extensive use of operant methods and ning with listener behavior and moving on to
experimental techniques but ironically draws the echoic and finally naming, appear to us
little from Skinner's interpretation of verbal to provide the necessary behavioral condi-
behavior. Part of the aim of the present paper tions for the acquisition of naming. But this
is to add to Skinner's exposition in ways that account, together with a range of related is-
will demonstrate the merits of a behavioral sues, needs to be experimentally validated. To
perspective. Although we have concentrated do so requires a systematic program of devel-
upon how naming, the most basic verbal unit, opmental behavioral research.
comes about during the first 2 years of life, There are a number of issues that such re-
our approach and method could be extended search should address. For example, we have
to account for the full range of linguistic be- indicated the primacy of listener behavior in
havior. the naming acquisition sequence (at least un-
Our main contribution, we believe, has til the establishment of higher order name
been to add to Skinner's description of speak- relations); this indeed appears to be generally
er behavior an account of the speaker as lis- true for young children who have no sensory
tener. As we have tried to show, the fusion of impairment, but there is evidence that chil-
speaker and listener behavior within the in- dren who do have sensory impairments (i.e.,
dividual has profound and far-reaching con- who are deaf or blind) may often learn speak-
sequences, transforming the basic verbal re- er behavior before they learn the correspond-
lations identified by Skinner (i.e., the tact, ing listener behavior relating to particular ob-
mand, and intraverbal) so that each becomes jects and events (Mulford, 1988; Pettito,
a variant of the basic name relation. The re- 1992). Whatever the particular sequence of
sult is the creation, within each name, of bi- acquisition, we maintain that for naming to
directional or closed-loop relations between a occur it is necessary for conventional listener
class of sometimes physically very different and speaker repertoires to combine, and we
objects or events and the speaker-listener be- have proposed that it is echoic responding,
havior it occasions. The name relation is both either overt or covert, in the presence of ob-
the focal point of this achievement, where all jects or events that is critical in bringing this
these events are brought together, and at the about. However, this possibility remains to be
same time the means of further dynamic in- empirically validated. Even in the case of chil-
terchange between them, giving rise to the dren without sensory impairments, it may be
range of emergent and symbolic behavior we informative in this regard to investigate forms
have described, including what has been of naming other than speaking, such as sign-
termed verbal thinking, reference, meaning, rule ing or other coding responses (see Lowen-
 19383711, 1996, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1901/jeab.1996.65-185 by Capes, Wiley Online Library on [26/05/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
NAMING AND OTHER SYMBOLIC BEHAVIOR 241
kron, 1984, 1988, 1989; Sundberg & Sund- the conditions that bring about this latter de-
berg, 1990). velopment, the better equipped will be the
It is also necessary to investigate the role researchers who embark upon these projects.
of reinforcement provided by the verbal com- The critical tests of whether success in estab-
munity in establishing and maintaining these lishing naming has been achieved are those
behavioral relations. The related issue of co- that we have already outlined. To date there
vert behavior and stimuli, including reinforc- is no convincing evidence that any nonhu-
ing stimuli, and the development of a covert man animal has passed tests of this kind.
verbal repertoire also require detailed re- The importance of studying verbal behav-
search because, as Skinner (1957, pp. 434- ior was made clear by Skinner (1987) who
438) has observed, such repertoires consti- wrote, "The human species took a crucial
tute so much of what is called human think- step forward when its vocal musculature came
ing. There are, however, complex method- under operant control in the production of
ological issues that must be dealt with if speech sounds. Indeed, it is possible that all
covert events are to be studied effectively in the distinctive achievements of the species
behavioral research (Lowe, 1983), not least in can be traced to that one genetic change" (p.
the study of stimulus equivalence. There are, 79). In our attempt to advance this study we
in addition, many issues concerning the role have focused upon the earliest forms of ver-
of naming in bringing about stimulus classes bal behavior, showing how children learn to
and functional transfer of behavior across name objects and events. In so doing, our ac-
stimulus classes. Because theories of stimulus count provides the prerequisite basis for an
equivalence that are not based on naming analysis of other more elaborate verbal forms,
have hitherto been dominant within behavior including complex propositions, sentences,
and grammar. We have also furnished a the-
analysis, research on the pivotal role of nam- oretical basis for understanding a range of in-
ing in such behavior has hardly begun. teractions between verbal and nonverbal be-
The present account also has implications havior that are of great practical import,
for attempts to teach language to nonhuman heretofore studied under the various guises
animals and to establish clear criteria against of rule-governed behavior, verbal control,
which such efforts can be assessed. It indi- correspondence training, and stimulus equiv-
cates that the central focus of such training alence (see Lowe, 1979, 1983; Lowe et al.,
should be upon establishing in the subjects a 1987). An understanding of the role of nam-
fusion of listener and speaker repertoires ing and other verbal behavior within all these
such as is established in the young child. domains represents an exciting challenge for
Clearly, the more that can be learned about future research.